diff --git a/.gitattributes b/.gitattributes index ae38eff66aaa40874cf224b236d6ebc42204faf3..7b86200cb34ec5d2fe983b6e387913e876bdd883 100644 --- a/.gitattributes +++ b/.gitattributes @@ -36,3 +36,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text RKD_Processing_line_Only/train.log filter=lfs diff=lfs merge=lfs -text VLM_Only/RKD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text VLM_Only/MGD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text +experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log filter=lfs diff=lfs merge=lfs -text diff --git a/experiment_cfg/processing_line_only/retrain_best/rkd_seed44_phase3_from_hf.log b/experiment_cfg/processing_line_only/retrain_best/rkd_seed44_phase3_from_hf.log new file mode 100644 index 0000000000000000000000000000000000000000..6241ec76b14d81af8313b974b8f73a0447fca13f --- /dev/null +++ b/experiment_cfg/processing_line_only/retrain_best/rkd_seed44_phase3_from_hf.log @@ -0,0 +1,3405 @@ +bash: line 1: python: command not found + +===== restart Fri Jun 12 07:46:05 PM KST 2026 ===== +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +***************************************** +Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +***************************************** +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +WARNING: mimicgen environments not imported since mimicgen is not installed! +WARNING: mimicgen environments not imported since mimicgen is not installed! +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/seonho/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/retrain_best/rkd_retrain_best2_phase3_only.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Using 2 GPUs + +================================================================================ +Starting sweep branch: rkd_temp_0.1 +Sweep vars: {'rkd_temp': 0.1} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase3_fm_rkd_fixed_0p5 +Policy type: groot_rkd +Base model path: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Output dir: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3 +Trainable preset: None +Policy overrides: {'rkd_enabled': True, 'rkd_fm_loss_weight': 1.0, 'rkd_loss_weight': 0.5, 'rkd_action_temp': 0.1, 'rkd_vlm_temp': 0.1, 'rkd_relation_mode': 'token_pair_mean'} +-------------------------------------------------------------------------------- + +[{'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustWaterTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustWaterTemperature', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffStove/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'TurnOffStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnStove/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustToasterOvenTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustToasterOvenTemperature', 'split': 'pretrain', 'source': 'human'}] +Using 100 subset demos for filter_key: 100_demos +/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/data/transform/state_action.py:257: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor). + self.statistics[key] = torch.tensor(value) +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/seonho/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/retrain_best/rkd_retrain_best2_phase3_only.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: TrueUsing 100 subset demos for filter_key: 100_demos + +ds_weights_alpha: 0.4 +================================================== + +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 2 GPUs +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos + +================================================================================ +Starting sweep branch: rkd_temp_0.1 +Sweep vars: {'rkd_temp': 0.1} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase3_fm_rkd_fixed_0p5 +Policy type: groot_rkd +Base model path: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Output dir: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3 +Trainable preset: None +Policy overrides: {'rkd_enabled': True, 'rkd_fm_loss_weight': 1.0, 'rkd_loss_weight': 0.5, 'rkd_action_temp': 0.1, 'rkd_vlm_temp': 0.1, 'rkd_relation_mode': 'token_pair_mean'} +-------------------------------------------------------------------------------- + +[{'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustWaterTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustWaterTemperature', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffStove/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'TurnOffStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnStove/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustToasterOvenTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustToasterOvenTemperature', 'split': 'pretrain', 'source': 'human'}] +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/data/transform/state_action.py:257: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor). + self.statistics[key] = torch.tensor(value) +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +dataset weights: [1. 0.88494669 0.76443286 0.83888607 0.73785464 1.00501644 + 0.80029956 0.66065525 0.83967491 0.83515002 0.95047759 0.86791692 + 0.88329477 0.78510653 0.64381843 0.67404766 0.69697218 0.60443684 + 0.78448103 0.83447486 0.61146215 0.64384058 0.79545025 0.93845856 + 0.75517122 0.7973985 ] +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Loaded 26 datasets +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Loading pretrained dual brain from /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Tune backbone vision tower: False +Tune backbone LLM: False +Tune action head projector: True +Tune action head DiT: True +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +dataset weights: [1. 0.88494669 0.76443286 0.83888607 0.73785464 1.00501644 + 0.80029956 0.66065525 0.83967491 0.83515002 0.95047759 0.86791692 + 0.88329477 0.78510653 0.64381843 0.67404766 0.69697218 0.60443684 + 0.78448103 0.83447486 0.61146215 0.64384058 0.79545025 0.93845856 + 0.75517122 0.7973985 ] +Loaded 26 datasets +Loading pretrained dual brain from /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Tune backbone vision tower: False +Tune backbone LLM: False +Tune action head projector: True +Tune action head DiT: True +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +Tune backbone llm: False +Tune backbone visual: True +Total number of DiT parameters: 550386688 +Tune backbone llm: False +Tune backbone visual: True +Total number of DiT parameters: 550386688 +Total number of SelfAttentionTransformer parameters: 201433088 +Tune action head projector: True +Tune action head diffusion model: True + Loading checkpoint shards: 0%| | 0/2 [00:00", line 198, in _run_module_as_main + File "", line 88, in _run_code + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 896, in + main() + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run + time.sleep(monitor_interval) + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler + raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval) +torch.distributed.elastic.multiprocessing.api.SignalException: Process 159527 got signal: 15 diff --git a/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log b/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log new file mode 100644 index 0000000000000000000000000000000000000000..0d17a98ae4ba7cb773fc32b35afc55af191353f9 --- /dev/null +++ b/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a245b8243ddcf5da916b9d670ccb3744c8659bb323295269fb0d607782ae851 +size 18113927 diff --git a/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu23_lane.log b/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu23_lane.log new file mode 100644 index 0000000000000000000000000000000000000000..48c648f239ea8e48a6434ecb787b508d6ca34795 --- /dev/null +++ b/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu23_lane.log @@ -0,0 +1,1076 @@ +===== TRAIN mask_ratio=0.4 devices=2,3 started at Mon Jun 15 12:09:43 PM KST 2026 ===== +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +***************************************** +Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +***************************************** +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +WARNING: mimicgen environments not imported since mimicgen is not installed! +WARNING: mimicgen environments not imported since mimicgen is not installed! +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/seonho/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation_0p4.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Using 2 GPUs + +================================================================================ +Starting sweep branch: mgd_token_mask_ratio_0.4 +Sweep vars: {'mgd_token_mask_ratio': 0.4} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase2_mgd_only +Policy type: groot_mgd +Base model path: /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Output dir: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.4/phase2 +Trainable preset: processing_line_only +Policy overrides: {'mgd_enabled': True, 'mgd_fm_loss_weight': 0.0, 'mgd_loss_weight': 1.0, 'mgd_sequence_hidden_dim': 512, 'mgd_token_mask_ratio': 0.4, 'mgd_use_cosine_loss': True, 'mgd_use_mse_loss': False} +-------------------------------------------------------------------------------- + +[{'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustWaterTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustWaterTemperature', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffStove/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'TurnOffStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnStove/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustToasterOvenTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustToasterOvenTemperature', 'split': 'pretrain', 'source': 'human'}] +Using 100 subset demos for filter_key: 100_demos +/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/data/transform/state_action.py:257: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor). + self.statistics[key] = torch.tensor(value) +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/seonho/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation_0p4.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 2 GPUs +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos + +================================================================================ +Starting sweep branch: mgd_token_mask_ratio_0.4 +Sweep vars: {'mgd_token_mask_ratio': 0.4} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase2_mgd_only +Policy type: groot_mgd +Base model path: /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Output dir: /home/seonho/groot_robocasa/robocasa_v2/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.4/phase2 +Trainable preset: processing_line_only +Policy overrides: {'mgd_enabled': True, 'mgd_fm_loss_weight': 0.0, 'mgd_loss_weight': 1.0, 'mgd_sequence_hidden_dim': 512, 'mgd_token_mask_ratio': 0.4, 'mgd_use_cosine_loss': True, 'mgd_use_mse_loss': False} +-------------------------------------------------------------------------------- + +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +[{'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustWaterTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustWaterTemperature', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffStove/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'TurnOffStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnStove/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustToasterOvenTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustToasterOvenTemperature', 'split': 'pretrain', 'source': 'human'}] +Using 100 subset demos for filter_key: 100_demos +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +dataset weights: [1. 0.88494669 0.76443286 0.83888607 0.73785464 1.00501644 + 0.80029956 0.66065525 0.83967491 0.83515002 0.95047759 0.86791692 + 0.88329477 0.78510653 0.64381843 0.67404766 0.69697218 0.60443684 + 0.78448103 0.83447486 0.61146215 0.64384058 0.79545025 0.93845856 + 0.75517122 0.7973985 ] +/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/data/transform/state_action.py:257: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor). + self.statistics[key] = torch.tensor(value) +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Loaded 26 datasets +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Loading pretrained dual brain from /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Tune backbone vision tower: False +Tune backbone LLM: False +Tune action head projector: False +Tune action head DiT: False +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +dataset weights: [1. 0.88494669 0.76443286 0.83888607 0.73785464 1.00501644 + 0.80029956 0.66065525 0.83967491 0.83515002 0.95047759 0.86791692 + 0.88329477 0.78510653 0.64381843 0.67404766 0.69697218 0.60443684 + 0.78448103 0.83447486 0.61146215 0.64384058 0.79545025 0.93845856 + 0.75517122 0.7973985 ] +Loaded 26 datasets +Loading pretrained dual brain from /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Tune backbone vision tower: False +Tune backbone LLM: False +Tune action head projector: False +Tune action head DiT: False +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/Atomic26_baseline/checkpoint-60000 +Tune backbone llm: False +Tune backbone visual: True +Total number of DiT parameters: 550386688 +Tune backbone llm: False +Tune backbone visual: True +Total number of DiT parameters: 550386688 +Total number of SelfAttentionTransformer parameters: 201433088 +Tune action head projector: True +Tune action head diffusion model: True + Loading checkpoint shards: 0%| | 0/2 [00:00", line 198, in _run_module_as_main + File "", line 88, in _run_code + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 896, in + main() + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run + time.sleep(monitor_interval) + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler + raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval) +torch.distributed.elastic.multiprocessing.api.SignalException: Process 159527 got signal: 15 +Terminated diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/config.json b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b894d895dfd93af5a9b611efa841e06ebc8aeb3f --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/config.json @@ -0,0 +1,73 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_RKD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "rkd_action_temp": 0.1, + "rkd_enabled": true, + "rkd_fm_loss_weight": 1.0, + "rkd_loss_weight": 0.5, + "rkd_loss_weight_end": 0.0, + "rkd_loss_weight_schedule": null, + "rkd_loss_weight_start": 0.1, + "rkd_relation_mode": "token_pair_mean", + "rkd_vlm_temp": 0.1, + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bb147cffa77f06a918deadacaa14b4b10d82c382 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d907a2ab38eeafc2144f692975a2ea8743fee421f16867eece069e40c72d6ba +size 4999367032 diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..de3af6eb825d0dcf96dbc28a675289fad65f93c8 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12aeab1fe2b8ad478d76dc85e59332af134f3867496c0bde7084f3496497fc8e +size 2586705312 diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model.safetensors.index.json b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/model.safetensors.index.json @@ -0,0 +1,906 @@ +{ + "metadata": { + "total_size": 7585939328 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors" + } +} diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/training_args.bin b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..767f6b0c6f1ab0c33e591d9ab26157ee00d20eea --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed18136a5e063abb170e8d790206843cf1594e52fb332edaee1e4510aa88ed25 +size 5905 diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/resolved_config.yaml b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/resolved_config.yaml index 7e7121ac995f77d9c8cbe3ed8096d5a5e97cbe99..165b67cfdfe504e43cc6952810afe7c4c0e3f18b 100644 --- a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/resolved_config.yaml +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/resolved_config.yaml @@ -1,6 +1,6 @@ -experiment_name: rkd_phase2_phase3_aux_fixed_0p5_retrain_seed44 +experiment_name: rkd_phase3_aux_fixed_0p5_retrain_seed44 policy_type: groot_rkd -base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-60000 +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44 sweep: rkd_temp: @@ -13,22 +13,6 @@ training: seed: 44 model: null phases: -- name: phase2_rkd_only - max_steps: 30000 - save_steps: 0 - trainable: - preset: processing_line_only - tune_llm: false - tune_visual: false - tune_projector: false - tune_diffusion_model: false - losses: - rkd_enabled: true - rkd_fm_loss_weight: 0.0 - rkd_loss_weight: 1.0 - rkd_action_temp: 0.1 - rkd_vlm_temp: 0.1 - rkd_relation_mode: token_pair_mean - name: phase3_fm_rkd_fixed_0p5 max_steps: 30000 save_steps: 0 diff --git a/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/rkd_retrain_best2_phase3_only.yaml b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/rkd_retrain_best2_phase3_only.yaml new file mode 100644 index 0000000000000000000000000000000000000000..486fd2354be830d6b9fe240ef6200ec4250862a0 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/rkd_retrain_best2_phase3_only.yaml @@ -0,0 +1,30 @@ +experiment_name: rkd_phase3_aux_fixed_0p5_retrain_seed44 +policy_type: groot_rkd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44/rkd_temp_0.1/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/retrain_best/rkd_seed44 +sweep: + rkd_temp: + - 0.1 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 + seed: 44 +model: null +phases: +- name: phase3_fm_rkd_fixed_0p5 + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + rkd_enabled: true + rkd_fm_loss_weight: 1.0 + rkd_loss_weight: 0.5 + rkd_action_temp: ${sweep.rkd_temp} + rkd_vlm_temp: ${sweep.rkd_temp} + rkd_relation_mode: token_pair_mean diff --git a/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/config.json b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b894d895dfd93af5a9b611efa841e06ebc8aeb3f --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/config.json @@ -0,0 +1,73 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_RKD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "rkd_action_temp": 0.1, + "rkd_enabled": true, + "rkd_fm_loss_weight": 1.0, + "rkd_loss_weight": 0.5, + "rkd_loss_weight_end": 0.0, + "rkd_loss_weight_schedule": null, + "rkd_loss_weight_start": 0.1, + "rkd_relation_mode": "token_pair_mean", + "rkd_vlm_temp": 0.1, + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9500f398c3dfc0777f95890e35d84e60b5e623ba --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a20a4468bc793dfe2673af8caaca55d77d3b51b184d2bf7efc87502b3d2feb66 +size 4999367032 diff --git a/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..84e2646202ca5394cc85b89e85e2bccb3dfca532 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3aaf86b6ddc18be35ff4a91b01930fb874f544c118ab380d1ae5d480aca419d4 +size 2586705312 diff --git a/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model.safetensors.index.json b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/model.safetensors.index.json @@ -0,0 +1,906 @@ +{ + "metadata": { + "total_size": 7585939328 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors" + } +} diff --git a/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/training_args.bin b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..10d8da1bc6f67dbc6e86765829d58e34306f970f --- /dev/null +++ b/processing_line_only/retrain_best/rkd_seed45/rkd_temp_0.1/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36e2c84120c47dcf3d3b3f57a98dbd057cbe716bb9e4a9870f70f8c6ef606a2e +size 5905 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c9509993e12e3bc3b291d1a0398318e256c8a2b7 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..28fdf1615eb9d0d602225715e34e523defff9783 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6ee1f00a02297a9f359b7af9710d808ca7208f22399cca90e557629359184da +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..ba241342be19ed217c3aaf5a6e782471d2707f98 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6d4617a50d34e2c44ef05433fea3c978b4f60fa0bbe0afb83e73e4a2ddd9607 +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..72933e0c23fd5dfa54588c6f2a4fd8573134866f --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7776b2ee616424ceee08de07a06227bb186644520814307016f9b550f46226d1 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ac02354431de894b3d1b42b3d8de7c1d1dfe12f9 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c236c8d4c3dcfbb24787ef44938f947fb2be5d973f70d9c0650b994b14b73113 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c17aeeb2347ccac2faebc6a887fed3e28c5671c0 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_cosine_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14e966cc29fc571161b938899aea45bb66cf7eff03db1f7263303a966ded550b +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/mgd_v2_loss_mse.yaml b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/mgd_v2_loss_mse.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1883fd04a65739d08ae48628c68e09cf0d0e9bd7 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/mgd_v2_loss_mse.yaml @@ -0,0 +1,47 @@ +experiment_name: mgd_v2_loss_mse +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-60000 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse +sweep: + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 + seed: 42 +model: null +phases: +- name: phase2_mgd_only + max_steps: 30000 + save_steps: 0 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false + losses: + mgd_enabled: true + mgd_fm_loss_weight: 0.0 + mgd_loss_weight: 1.0 + mgd_sequence_hidden_dim: 512 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: false + mgd_use_mse_loss: true +- name: phase3_fm_mgd_fixed_0p5 + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: true + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.5 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: false + mgd_use_mse_loss: true diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/config.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6dd26df6e81cab44507bbf62a5f2ab994e82673c --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": true, + "mgd_fm_loss_weight": 0.0, + "mgd_loss_weight": 1.0, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": null, + "mgd_loss_weight_start": 0.05, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.3, + "mgd_use_cosine_loss": false, + "mgd_use_mse_loss": true, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/experiment_cfg/metadata.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c9509993e12e3bc3b291d1a0398318e256c8a2b7 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f4f0dbc5f793a4ae794e2f96906655d4859e9a4b --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d983182f2aeca91d08f6fcbb64f4a695f3d4b87e411c082768897249062d8281 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model.safetensors.index.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/trainer_state.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..a3841a47db9e667e7c89f0ac68a69ea05ecea858 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.2993220746517181, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.016896762326359747, + "grad_norm": 0.005162701942026615, + "kept_token_count": 573.521875, + "learning_rate": 1.8e-07, + "loss": 0.0039, + "mgd_cos_sim": -0.008513879729434848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003939507296308875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.411997604370114, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6078704833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.30044272541999817, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.01597237167879939, + "grad_norm": 0.005462351720780134, + "kept_token_count": 572.4703125, + "learning_rate": 3.8e-07, + "loss": 0.0039, + "mgd_cos_sim": -0.004833314369898289, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003925130167044699, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.9530834197998, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8565155029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.29848135709762574, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.01795398611575365, + "grad_norm": 0.0053020683117210865, + "kept_token_count": 574.10625, + "learning_rate": 5.800000000000001e-07, + "loss": 0.0039, + "mgd_cos_sim": -3.2724009361118076e-05, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003906377800740301, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.739921188354494, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72167053222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.29967315793037413, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.013898858055472373, + "grad_norm": 0.005095999222248793, + "kept_token_count": 573.23125, + "learning_rate": 7.799999999999999e-07, + "loss": 0.0039, + "mgd_cos_sim": 0.008340401132591069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0038736702874302862, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.99527206420898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5004486083984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 40, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.30030018389225005, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.014851205237209796, + "grad_norm": 0.005439987871795893, + "kept_token_count": 572.5328125, + "learning_rate": 9.8e-07, + "loss": 0.0038, + "mgd_cos_sim": 0.02362883957102895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003813949855975807, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.77116928100586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2824310302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 50, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.3002846360206604, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.016715700179338454, + "grad_norm": 0.005381753668189049, + "kept_token_count": 572.4578125, + "learning_rate": 1.18e-06, + "loss": 0.0038, + "mgd_cos_sim": 0.038696485199034215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0037550919223576784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.51955108642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.62437133789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 60, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30021377801895144, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.018978586327284573, + "grad_norm": 0.00545977707952261, + "kept_token_count": 572.70625, + "learning_rate": 1.38e-06, + "loss": 0.0037, + "mgd_cos_sim": 0.055935827642679216, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003687750664539635, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.3760772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8497375488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 70, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3006816476583481, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.015217910427600146, + "grad_norm": 0.005412853788584471, + "kept_token_count": 572.246875, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0036, + "mgd_cos_sim": 0.08307901173830032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0035817226395010947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.51319885253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0600555419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 80, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3005389392375946, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.01751301409676671, + "grad_norm": 0.00560818100348115, + "kept_token_count": 572.2484375, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0035, + "mgd_cos_sim": 0.10970342978835106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003477721009403467, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.86912689208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.85289306640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 90, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29997555911540985, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.016089385095983743, + "grad_norm": 0.005010072607547045, + "kept_token_count": 572.8796875, + "learning_rate": 1.98e-06, + "loss": 0.0034, + "mgd_cos_sim": 0.1404570683836937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0033575895708054305, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.43016357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.06866455078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 100, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30113585293293, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.018978271540254353, + "grad_norm": 0.0050217825919389725, + "kept_token_count": 571.7328125, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0032, + "mgd_cos_sim": 0.179158815741539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.003206410934217274, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.57443695068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3827819824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 110, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2995553553104401, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.016047739516943695, + "grad_norm": 0.004885856993496418, + "kept_token_count": 573.084375, + "learning_rate": 2.38e-06, + "loss": 0.0031, + "mgd_cos_sim": 0.21575351506471635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0030634628608822824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.60358428955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63617553710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 120, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3012469410896301, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.017259843368083238, + "grad_norm": 0.00463787792250514, + "kept_token_count": 571.5140625, + "learning_rate": 2.58e-06, + "loss": 0.0029, + "mgd_cos_sim": 0.25103116035461426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0029256595065817238, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.17133026123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5993377685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 130, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.29984443783760073, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.018337335018441082, + "grad_norm": 0.004889979027211666, + "kept_token_count": 572.971875, + "learning_rate": 2.78e-06, + "loss": 0.0028, + "mgd_cos_sim": 0.28638008832931516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.002787577826529741, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.87248001098632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8415008544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2992469877004623, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.01908568162471056, + "grad_norm": 0.004397881217300892, + "kept_token_count": 573.353125, + "learning_rate": 2.98e-06, + "loss": 0.0026, + "mgd_cos_sim": 0.3267642676830292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0026298271492123603, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.44259414672851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0873718261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 150, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2996245324611664, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.01722847083583474, + "grad_norm": 0.004124379716813564, + "kept_token_count": 573.1484375, + "learning_rate": 3.18e-06, + "loss": 0.0025, + "mgd_cos_sim": 0.3621167421340942, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0024917315458878875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.77808074951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8490234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 160, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.299912703037262, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.015844112448394298, + "grad_norm": 0.004050443414598703, + "kept_token_count": 572.65625, + "learning_rate": 3.38e-06, + "loss": 0.0023, + "mgd_cos_sim": 0.4012854814529419, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0023387286579236387, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.85009994506837, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4463134765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.30002049207687376, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.014265630580484866, + "grad_norm": 0.003723439760506153, + "kept_token_count": 572.8859375, + "learning_rate": 3.58e-06, + "loss": 0.0022, + "mgd_cos_sim": 0.4371758043766022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0021985321305692195, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.79822082519532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0907928466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 180, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3000711590051651, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.015206280350685119, + "grad_norm": 0.0036274422891438007, + "kept_token_count": 572.8671875, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0021, + "mgd_cos_sim": 0.4731670439243317, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0020579413045197724, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 103.03926467895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.27357788085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 190, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.300471892952919, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.016573692113161086, + "grad_norm": 0.00349211017601192, + "kept_token_count": 572.35625, + "learning_rate": 3.98e-06, + "loss": 0.0019, + "mgd_cos_sim": 0.5137595146894455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0018993770121596754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.82919158935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.00258483886716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 200, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30000542402267455, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.017103804741054773, + "grad_norm": 0.003404842223972082, + "kept_token_count": 573.0140625, + "learning_rate": 4.18e-06, + "loss": 0.0018, + "mgd_cos_sim": 0.5443079948425293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0017800470814108849, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.29113235473633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.01099548339846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 210, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.29939305782318115, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.0156393731944263, + "grad_norm": 0.0033902411814779043, + "kept_token_count": 572.9890625, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0017, + "mgd_cos_sim": 0.5685996174812317, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.001685157825704664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 117.51760940551758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08617248535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 220, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.30042233467102053, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.019152250699698926, + "grad_norm": 0.003287574043497443, + "kept_token_count": 572.1953125, + "learning_rate": 4.58e-06, + "loss": 0.0016, + "mgd_cos_sim": 0.5927348077297211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0015908797038719058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.0850730895996, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2318939208984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 230, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.2993056267499924, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.02037977511063218, + "grad_norm": 0.0030089709907770157, + "kept_token_count": 573.1453125, + "learning_rate": 4.78e-06, + "loss": 0.0015, + "mgd_cos_sim": 0.6243551015853882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0014673630357719958, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.50991973876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.89459228515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 240, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.30110191404819486, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.01756344409659505, + "grad_norm": 0.0030740518122911453, + "kept_token_count": 572.0875, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0014, + "mgd_cos_sim": 0.6453717112541199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0013852669042535126, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.82322540283204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61517944335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 250, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.30005758702754975, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.017216639872640372, + "grad_norm": 0.002936710836365819, + "kept_token_count": 572.7515625, + "learning_rate": 5.18e-06, + "loss": 0.0013, + "mgd_cos_sim": 0.6689373254776001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0012932137004099786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.16112823486327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5552062988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 260, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.300126326084137, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.019506706111133098, + "grad_norm": 0.002778518246486783, + "kept_token_count": 572.575, + "learning_rate": 5.38e-06, + "loss": 0.0012, + "mgd_cos_sim": 0.6895586490631104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.001212661643512547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.20398864746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.86768798828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 270, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3003015607595444, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.017091846093535423, + "grad_norm": 0.0026404678355902433, + "kept_token_count": 572.4359375, + "learning_rate": 5.58e-06, + "loss": 0.0011, + "mgd_cos_sim": 0.7080845057964325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0011402950040064752, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.25829620361327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.64854125976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 280, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.299298033118248, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.017366492282599212, + "grad_norm": 0.0024456954561173916, + "kept_token_count": 573.2578125, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.001, + "mgd_cos_sim": 0.7323636829853057, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.001045454468112439, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.01609649658204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8513244628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 290, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29960326850414276, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.021023061871528626, + "grad_norm": 0.0022826327476650476, + "kept_token_count": 572.9140625, + "learning_rate": 5.98e-06, + "loss": 0.001, + "mgd_cos_sim": 0.7435183167457581, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.001001881738193333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.46307983398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28957214355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 300, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.3001630365848541, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.021500048320740463, + "grad_norm": 0.002912576077505946, + "kept_token_count": 572.709375, + "learning_rate": 6.18e-06, + "loss": 0.0009, + "mgd_cos_sim": 0.7593915939331055, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0009398767782840878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.45773620605468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6007354736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.299407097697258, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.015855793934315444, + "grad_norm": 0.0020337949972599745, + "kept_token_count": 573.2296875, + "learning_rate": 6.38e-06, + "loss": 0.0009, + "mgd_cos_sim": 0.7744471848011016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0008810658648144454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.18916931152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34499206542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 320, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.3002766162157059, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.02047953028231859, + "grad_norm": 0.0019079569028690457, + "kept_token_count": 572.459375, + "learning_rate": 6.58e-06, + "loss": 0.0008, + "mgd_cos_sim": 0.7879986703395844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0008281303627882153, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.79286499023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.86212158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 330, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2999196231365204, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.025936092995107175, + "grad_norm": 0.0017581601860001683, + "kept_token_count": 572.8859375, + "learning_rate": 6.78e-06, + "loss": 0.0008, + "mgd_cos_sim": 0.7999847948551178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000781309581361711, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.0246841430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.04309997558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 340, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.29987775087356566, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.021336471661925316, + "grad_norm": 0.0016862057382240891, + "kept_token_count": 572.68125, + "learning_rate": 6.98e-06, + "loss": 0.0007, + "mgd_cos_sim": 0.8116973519325257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0007355574343819171, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.49362335205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3100921630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 350, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2986245483160019, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.025641509145498276, + "grad_norm": 0.0014733916614204645, + "kept_token_count": 574.0890625, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0007, + "mgd_cos_sim": 0.8165015280246735, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0007167910516727716, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.50485229492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8868408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 360, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.30021290183067323, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.023436378967016936, + "grad_norm": 0.0015670913271605968, + "kept_token_count": 572.728125, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0007, + "mgd_cos_sim": 0.8237797319889069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0006883606081828475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.54068450927736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.597412109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 370, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29927953481674197, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.0227240277454257, + "grad_norm": 0.0013380766613408923, + "kept_token_count": 573.3328125, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8339468955993652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0006486451253294945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.03687286376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.19397888183596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 380, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30013754665851594, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.023633824475109578, + "grad_norm": 0.0012328452430665493, + "kept_token_count": 572.55, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8471628248691558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005970203725155443, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.6213409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.8713653564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 390, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.30093667805194857, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.0251630375161767, + "grad_norm": 0.001068549114279449, + "kept_token_count": 571.96875, + "learning_rate": 7.98e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8449893295764923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0006055105826817453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.9017318725586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7976867675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 400, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30021707713603973, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.028849288634955884, + "grad_norm": 0.0010010290425270796, + "kept_token_count": 572.44375, + "learning_rate": 8.18e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8522169649600982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005772776436060667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.20013122558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1985321044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 410, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.2998775660991669, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.02690453752875328, + "grad_norm": 0.0010898783802986145, + "kept_token_count": 572.896875, + "learning_rate": 8.38e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8531998455524444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005734382837545126, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.08846893310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5649353027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 420, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3008458405733109, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.032312828302383426, + "grad_norm": 0.0009225759422406554, + "kept_token_count": 572.01875, + "learning_rate": 8.58e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.852482146024704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005762418149970472, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.79128723144532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.21300354003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2997849613428116, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.028885131143033503, + "grad_norm": 0.0007574241026304662, + "kept_token_count": 573.165625, + "learning_rate": 8.78e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8524056792259216, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005765405134297907, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.09068298339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.03685302734374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 440, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.29949448108673093, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.030041805282235146, + "grad_norm": 0.0007696865359321237, + "kept_token_count": 573.125, + "learning_rate": 8.98e-06, + "loss": 0.0005, + "mgd_cos_sim": 0.8620670914649964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005388005927670747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.64935302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2593231201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2997928947210312, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.03078779298812151, + "grad_norm": 0.0007006721571087837, + "kept_token_count": 572.9515625, + "learning_rate": 9.18e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8591503977775574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005501939449459314, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.68341064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.60784606933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 460, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30027629137039186, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.032908044010400775, + "grad_norm": 0.0005534766823984683, + "kept_token_count": 572.615625, + "learning_rate": 9.38e-06, + "loss": 0.0006, + "mgd_cos_sim": 0.8546763777732849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000567670539021492, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 206.56573181152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.33544006347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 470, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2998074650764465, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.03405827563256025, + "grad_norm": 0.0006789539474993944, + "kept_token_count": 573.05, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0005, + "mgd_cos_sim": 0.8663884222507476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005219204089371487, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.92254943847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.64561767578124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 480, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2989975869655609, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.0361887464299798, + "grad_norm": 0.0006759435636922717, + "kept_token_count": 573.703125, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0005, + "mgd_cos_sim": 0.8630551636219025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005349409009795636, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.64740753173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.14403381347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 490, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3003123939037323, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.03399752154946327, + "grad_norm": 0.0006413013907149434, + "kept_token_count": 572.3234375, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0005, + "mgd_cos_sim": 0.8659166276454926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000523763382807374, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.3918670654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1339935302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 500, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.30019292831420896, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.03350156042724848, + "grad_norm": 0.0013001980260014534, + "kept_token_count": 572.5796875, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8696651875972747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005091204890049994, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.31466217041014, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.88263244628905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 510, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.300015589594841, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.04141273219138384, + "grad_norm": 0.0005477690719999373, + "kept_token_count": 572.846875, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.0006, + "mgd_cos_sim": 0.857114839553833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005581452860496938, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 183.7188720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8968963623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 520, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.3010154694318771, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.04092707280069589, + "grad_norm": 0.0009266670676879585, + "kept_token_count": 571.9109375, + "learning_rate": 1.058e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8677377343177796, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005166496179299429, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.57647857666015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71034545898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 530, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3000517517328262, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.04723755456507206, + "grad_norm": 0.0006457074196077883, + "kept_token_count": 572.6796875, + "learning_rate": 1.078e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8644467234611511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005295051843859256, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.3512420654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.84951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 540, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.3004267692565918, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.04228602796792984, + "grad_norm": 0.0007618220406584442, + "kept_token_count": 572.6375, + "learning_rate": 1.098e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8647011458873749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005285113584250212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.3516387939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2547576904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 550, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.30095221996307375, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.050832181423902514, + "grad_norm": 0.0006954015698283911, + "kept_token_count": 571.86875, + "learning_rate": 1.118e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8670834302902222, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005192055250518024, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 125.64144592285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4080474853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 560, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.3001033991575241, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.05408768840134144, + "grad_norm": 0.001220403122715652, + "kept_token_count": 572.653125, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8659987032413483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005234427400864661, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.48141937255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5552062988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 570, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.29988465309143064, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.05566973276436329, + "grad_norm": 0.0008535711094737053, + "kept_token_count": 572.721875, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8683864355087281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005141157249454409, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 102.00904388427735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21641845703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 580, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.29988857805728913, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.06312111467123031, + "grad_norm": 0.0008355358149856329, + "kept_token_count": 572.8671875, + "learning_rate": 1.178e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.869775265455246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005086905031930656, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.63275833129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1623565673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 590, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2988591820001602, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.06672325320541858, + "grad_norm": 0.0005324167432263494, + "kept_token_count": 573.7796875, + "learning_rate": 1.198e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8699969232082367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000507824623491615, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.27859725952149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5270263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 600, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29956924021244047, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.07103580310940742, + "grad_norm": 0.00038888142444193363, + "kept_token_count": 573.0421875, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8779703438282013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004766784957610071, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.30781402587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.0099639892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 610, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2997678190469742, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.0792984675616026, + "grad_norm": 0.00045262216008268297, + "kept_token_count": 572.8265625, + "learning_rate": 1.238e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8747454285621643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004892758326604963, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.78878479003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3571838378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 620, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.29975218772888185, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.0859006479382515, + "grad_norm": 0.0005028633750043809, + "kept_token_count": 573.071875, + "learning_rate": 1.258e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8748368799686432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004889185860520229, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.14962387084961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.07470703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 630, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.29929859936237335, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.08705438524484635, + "grad_norm": 0.00033351322053931653, + "kept_token_count": 573.3015625, + "learning_rate": 1.278e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.872400164604187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004984370898455381, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.82319488525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5390197753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 640, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3008156806230545, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.09534797593951225, + "grad_norm": 0.0005323089426383376, + "kept_token_count": 572.0140625, + "learning_rate": 1.298e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8710298836231232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005037897295551375, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.89800338745117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.96116943359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 650, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2999496728181839, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.0959438480436802, + "grad_norm": 0.0006515885470435023, + "kept_token_count": 572.7890625, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8706766247749329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0005051696847658604, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.23606491088867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.88884887695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 660, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2999166429042816, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.08730020821094513, + "grad_norm": 0.0005075734807178378, + "kept_token_count": 572.9375, + "learning_rate": 1.338e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8754428625106812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004865514812991023, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.29482879638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.88763122558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 670, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3008765369653702, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.09167036563158035, + "grad_norm": 0.0005102980649098754, + "kept_token_count": 572.025, + "learning_rate": 1.358e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8744859874248505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004902892571408302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.31705245971679, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8744873046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 680, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29990856945514677, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.09574926048517227, + "grad_norm": 0.0006365245790220797, + "kept_token_count": 572.928125, + "learning_rate": 1.378e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8740619599819184, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004919455968774855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.36969604492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3369445800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 690, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.3003203451633453, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.10177874565124512, + "grad_norm": 0.00043859879951924086, + "kept_token_count": 572.6984375, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8720228612422943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004999108263291418, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.34993896484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.06017150878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 700, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.29959323406219485, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.09537224471569061, + "grad_norm": 0.0005116482498124242, + "kept_token_count": 573.35, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8779477953910828, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00047676666581537576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.8557975769043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4831085205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 710, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.2989736944437027, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.09108255058526993, + "grad_norm": 0.0005484246648848057, + "kept_token_count": 573.45, + "learning_rate": 1.438e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8760976076126099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004839938454097137, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.79929428100586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.09300842285154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 720, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.2992690771818161, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.0918341375887394, + "grad_norm": 0.00041597578092478216, + "kept_token_count": 573.446875, + "learning_rate": 1.458e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8787913262844086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004734716669190675, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.04025344848633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.08448181152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 730, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.29924718737602235, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.10044587552547454, + "grad_norm": 0.00039654556894674897, + "kept_token_count": 573.3578125, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8773977518081665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00047891524736769496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.27124938964843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6502258300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 740, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3002506285905838, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.1021213561296463, + "grad_norm": 0.0005329240811988711, + "kept_token_count": 572.6703125, + "learning_rate": 1.498e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8799479246139527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004689536290243268, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.94109649658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.00218811035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 750, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30058678686618806, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.10231394916772843, + "grad_norm": 0.00044822078780271113, + "kept_token_count": 572.2171875, + "learning_rate": 1.518e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8808114230632782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004655805561924353, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.03971939086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3347961425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 760, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.3011605262756348, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.1125958390533924, + "grad_norm": 0.0004417009185999632, + "kept_token_count": 571.6875, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8727723777294158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000496983079938218, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.42799377441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.30845336914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 770, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.29936744570732116, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.1047093965113163, + "grad_norm": 0.0005664906930178404, + "kept_token_count": 573.171875, + "learning_rate": 1.558e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8807245790958405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004659197962610051, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.41718368530273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6737854003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 780, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.30152023434638975, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.11382413730025291, + "grad_norm": 0.0005546245956793427, + "kept_token_count": 571.221875, + "learning_rate": 1.578e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8823597729206085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00045953223889227957, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.99727630615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8636535644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 790, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.2996055901050568, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.11636945083737374, + "grad_norm": 0.000513254723045975, + "kept_token_count": 572.8953125, + "learning_rate": 1.598e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8820119559764862, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004608910356182605, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.5888069152832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1564514160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 800, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.2998342514038086, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.11115564927458763, + "grad_norm": 0.00045695898006670177, + "kept_token_count": 572.85, + "learning_rate": 1.618e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8834051787853241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004554487066343427, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.78758850097657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4174438476563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 810, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29947128891944885, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.12256432846188545, + "grad_norm": 0.0005765419919043779, + "kept_token_count": 573.078125, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0005, + "mgd_cos_sim": 0.8822998940944672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004597661434672773, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.58507614135742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0896697998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 820, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30041698515415194, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.12356646209955216, + "grad_norm": 0.00042899540858343244, + "kept_token_count": 572.2328125, + "learning_rate": 1.658e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8878526329994202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004380757629405707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.53077774047851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.74619750976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 830, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.30148373246192933, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.1240833654999733, + "grad_norm": 0.0005187083152122796, + "kept_token_count": 571.6, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8850381433963775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000449069932801649, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.51522903442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.65440673828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 840, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.29912412762641905, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.12314706593751908, + "grad_norm": 0.0006232348969206214, + "kept_token_count": 573.446875, + "learning_rate": 1.698e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8869918823242188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004414381139213219, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.80411987304687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.15009155273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 850, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2993906795978546, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.12161291316151619, + "grad_norm": 0.0007599382079206407, + "kept_token_count": 573.48125, + "learning_rate": 1.718e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8959633231163024, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00040639338549226524, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.37839202880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.55328369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 860, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.30000796020030973, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.1259131908416748, + "grad_norm": 0.0005116216489113867, + "kept_token_count": 572.540625, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8922711968421936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00042081575375050304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.32231979370117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.05312194824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 870, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.3007690817117691, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.13217667788267135, + "grad_norm": 0.00045276639866642654, + "kept_token_count": 572.35625, + "learning_rate": 1.758e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8918128788471222, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004226061195367947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.90813369750977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1158874511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 880, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.3007063359022141, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.12106014788150787, + "grad_norm": 0.0006305345450527966, + "kept_token_count": 572.0375, + "learning_rate": 1.778e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8956561505794525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004075934033608064, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.04292068481445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82398681640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 890, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2997522741556168, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.11951769143342972, + "grad_norm": 0.0005155567196197808, + "kept_token_count": 572.9171875, + "learning_rate": 1.798e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8978616952896118, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003989778662798926, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.49839324951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.63628845214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 900, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2994367957115173, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.11793501228094101, + "grad_norm": 0.0006188174011185765, + "kept_token_count": 573.115625, + "learning_rate": 1.818e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8964950799942016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004043162742163986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.8025016784668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9922302246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 910, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2997718721628189, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.12149952873587608, + "grad_norm": 0.000477587862405926, + "kept_token_count": 573.05, + "learning_rate": 1.838e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9001580536365509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00039000779506750407, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.29534759521485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5048126220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 920, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2997481465339661, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.12143412828445435, + "grad_norm": 0.0005622464814223349, + "kept_token_count": 573.103125, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8974729061126709, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000400496696238406, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.49475326538087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63880920410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 930, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.29984969198703765, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.11645803600549698, + "grad_norm": 0.000551174976862967, + "kept_token_count": 572.9140625, + "learning_rate": 1.878e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8971357822418213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00040181348158512263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.18343353271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44927978515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 940, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30046744644641876, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.1092892199754715, + "grad_norm": 0.0004586748837027699, + "kept_token_count": 572.6859375, + "learning_rate": 1.898e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9014615416526794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00038491601299028844, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.7420265197754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4782745361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 950, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.3003670275211334, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.11382054388523102, + "grad_norm": 0.00048785904073156416, + "kept_token_count": 572.6828125, + "learning_rate": 1.918e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9048151135444641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00037181610241532326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.38802719116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.90855102539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 960, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.2993121087551117, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.12087838351726532, + "grad_norm": 0.0004834515566471964, + "kept_token_count": 573.3546875, + "learning_rate": 1.938e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9003195524215698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003893768967827782, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.17539978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7879211425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2997093886137009, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.10373183414340019, + "grad_norm": 0.0006460681324824691, + "kept_token_count": 572.8875, + "learning_rate": 1.958e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9048317968845367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00037175098550505935, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.39765853881836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5056488037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 980, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.3013992100954056, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.11943678930401802, + "grad_norm": 0.0006123176426626742, + "kept_token_count": 571.8875, + "learning_rate": 1.978e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.8973571836948395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0004009486932773143, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.6933204650879, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.48955993652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 990, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.3001940965652466, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.11249091550707817, + "grad_norm": 0.0005046462756581604, + "kept_token_count": 572.68125, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9012540936470032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00038572638004552573, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.85162200927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9755554199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1000, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29978465735912324, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.10537007227540016, + "grad_norm": 0.0005588244530372322, + "kept_token_count": 572.9734375, + "learning_rate": 2.018e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9065454363822937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003650570783065632, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.83382034301758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79419860839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1010, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3007091462612152, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.10745846927165985, + "grad_norm": 0.0005992756341584027, + "kept_token_count": 572.1671875, + "learning_rate": 2.038e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9089935600757599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00035549410386011006, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.5470947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3565612792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2997182130813599, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.10269606858491898, + "grad_norm": 0.0004864966031163931, + "kept_token_count": 573.146875, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9137841403484345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003367809084011242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.19365539550782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.64447326660155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1030, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.299693101644516, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.10661713406443596, + "grad_norm": 0.0005320529453456402, + "kept_token_count": 573.1375, + "learning_rate": 2.078e-05, + "loss": 0.0004, + "mgd_cos_sim": 0.9085926651954651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00035706008493434637, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.14926223754883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8368194580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1040, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.299668949842453, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.09919600784778596, + "grad_norm": 0.0006303574773482978, + "kept_token_count": 573.353125, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9105334877967834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003494787495583296, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.12483444213868, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8539611816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1050, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.2997568428516388, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.09956851825118065, + "grad_norm": 0.0004959927755407989, + "kept_token_count": 572.89375, + "learning_rate": 2.118e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.914968067407608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003321561496704817, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.70266265869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0489776611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1060, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30054019689559935, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.11004777252674103, + "grad_norm": 0.0005873654154129326, + "kept_token_count": 572.471875, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9159470915794372, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00032833185978233816, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.53961029052735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8487060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1070, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.30030580759048464, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.11268563121557236, + "grad_norm": 0.00046965875662863255, + "kept_token_count": 572.5890625, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9137315571308136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00033698625629767774, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.16431579589843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.06675720214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1080, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29928335547447205, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.10737645477056504, + "grad_norm": 0.00048536184476688504, + "kept_token_count": 573.3515625, + "learning_rate": 2.178e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9131826639175415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003391304257092997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.07879638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36693115234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1090, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2989618510007858, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.09973540082573891, + "grad_norm": 0.0005106697208248079, + "kept_token_count": 573.596875, + "learning_rate": 2.198e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9143308997154236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003346450685057789, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.87580261230468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.89178466796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1100, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.300348699092865, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.10569946989417076, + "grad_norm": 0.0005278848111629486, + "kept_token_count": 572.5453125, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9178339898586273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003209611139027402, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.8284782409668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0844390869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1110, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2997534364461899, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.10557314306497574, + "grad_norm": 0.0005295761511661112, + "kept_token_count": 573.096875, + "learning_rate": 2.238e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9185760319232941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003180625644745305, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.85163803100586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5733612060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1120, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2996536433696747, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.11668928563594819, + "grad_norm": 0.0007156577776186168, + "kept_token_count": 573.1421875, + "learning_rate": 2.258e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9186311066150665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003178474114974961, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.93986892700195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0663208007812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1130, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2998663604259491, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.1226020723581314, + "grad_norm": 0.0005338524933904409, + "kept_token_count": 573.0734375, + "learning_rate": 2.278e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9162634074687958, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00032709623337723315, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.53534393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7358642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1140, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.2988866299390793, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.10601783618330955, + "grad_norm": 0.0006951354444026947, + "kept_token_count": 573.596875, + "learning_rate": 2.298e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9189613401889801, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003165574336890131, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.86083908081055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6227966308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1150, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29948090612888334, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.10726153962314129, + "grad_norm": 0.0006378886173479259, + "kept_token_count": 573.24375, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9223665237426758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00030325587722472845, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.75122528076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7123077392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1160, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2995778858661652, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.11589534059166909, + "grad_norm": 0.0006198103656060994, + "kept_token_count": 572.9046875, + "learning_rate": 2.338e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.923201709985733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00029999354446772484, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.44173126220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2938507080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1170, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.29956882297992704, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.11577474400401115, + "grad_norm": 0.0006296385545283556, + "kept_token_count": 573.25, + "learning_rate": 2.358e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9215001165866852, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0003066403849516064, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.8071388244629, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1794189453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1180, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.29951764047145846, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.10217884778976441, + "grad_norm": 0.0006928136572241783, + "kept_token_count": 573.2390625, + "learning_rate": 2.378e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9249614477157593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002931195456767455, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.360009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.92730407714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1190, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2990258127450943, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.11620615124702453, + "grad_norm": 0.0006918617873452604, + "kept_token_count": 573.7328125, + "learning_rate": 2.398e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9258788764476776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002895358455134556, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.8321304321289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0622253417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1200, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3000535398721695, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.10855448916554451, + "grad_norm": 0.0005144812748767436, + "kept_token_count": 572.496875, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9267842590808868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00028599913057405504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.63898315429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.25059814453124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1210, + "valid_token_count": 817.9078125 + }, + { + "actual_token_mask_ratio": 0.3003967344760895, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.1159663513302803, + "grad_norm": 0.000708865700289607, + "kept_token_count": 572.4484375, + "learning_rate": 2.438e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.929000836610794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002773405984044075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.10074996948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.57826232910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1220, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2996940791606903, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.11140702739357948, + "grad_norm": 0.0007636223454028368, + "kept_token_count": 572.6765625, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9280442953109741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002810771227814257, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.64011840820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8669128417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1230, + "valid_token_count": 817.75625 + }, + { + "actual_token_mask_ratio": 0.3000982701778412, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.10695070996880532, + "grad_norm": 0.0006223359378054738, + "kept_token_count": 572.703125, + "learning_rate": 2.478e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9282683849334716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00028020179888699204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.0314109802246, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3162048339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1240, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2998576879501343, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.11151221320033074, + "grad_norm": 0.00068870052928105, + "kept_token_count": 572.8390625, + "learning_rate": 2.498e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9333214461803436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00026046331913676113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.24414978027343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2460540771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1250, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.3000798851251602, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.11267503798007965, + "grad_norm": 0.0006094175041653216, + "kept_token_count": 572.821875, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.9358916282653809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00025042352062882857, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90674209594727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.85951843261716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1260, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2997191220521927, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.11182903945446014, + "grad_norm": 0.0005916907102800906, + "kept_token_count": 572.9890625, + "learning_rate": 2.538e-05, + "loss": 0.0003, + "mgd_cos_sim": 0.933410769701004, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00026011435402324424, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.20872421264649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.217724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1270, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.2989955484867096, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.11281263455748558, + "grad_norm": 0.0005631814710795879, + "kept_token_count": 573.6234375, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9395904839038849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00023597488325322047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.84120864868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5494842529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1280, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2999913215637207, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.11304367259144783, + "grad_norm": 0.00047603907296434045, + "kept_token_count": 572.5421875, + "learning_rate": 2.578e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9362745404243469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002489277423592284, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.4475082397461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3978637695312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1290, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.29987236559391023, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.11069310456514359, + "grad_norm": 0.0008984367595985532, + "kept_token_count": 572.7734375, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9381960570812226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00024142188631230964, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.91014251708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5271911621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1300, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.3004598766565323, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.1057704396545887, + "grad_norm": 0.0006301250541582704, + "kept_token_count": 572.4671875, + "learning_rate": 2.618e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9382399678230285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00024125033378368242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.76261672973632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5983001708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3008446365594864, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.09547134935855865, + "grad_norm": 0.0004464749654289335, + "kept_token_count": 572.0859375, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9437228560447692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00021983284241287038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7276237487793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.9212677001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1320, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30042627453804016, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.11258628070354462, + "grad_norm": 0.000533134734723717, + "kept_token_count": 572.5609375, + "learning_rate": 2.658e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9396860837936402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00023560149274999277, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.91520538330079, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1675323486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1330, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2993124067783356, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.10361904129385949, + "grad_norm": 0.00045692198909819126, + "kept_token_count": 573.5703125, + "learning_rate": 2.678e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9445885479450226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00021645116212312132, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10491180419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1118133544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1340, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.29916271567344666, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.11487215906381607, + "grad_norm": 0.0005396396736614406, + "kept_token_count": 573.478125, + "learning_rate": 2.698e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9431836843490601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002219389338279143, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96125183105468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.808154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1350, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3000751197338104, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.11391562819480897, + "grad_norm": 0.0006954174605198205, + "kept_token_count": 572.6875, + "learning_rate": 2.718e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9433498203754425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00022128995187813415, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.66007843017579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7853515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1360, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3002630710601807, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.1175438940525055, + "grad_norm": 0.0005821395316161215, + "kept_token_count": 572.6703125, + "learning_rate": 2.738e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9427393674850464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0002236744883703068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.1390769958496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.15513000488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1370, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30047530829906466, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.10228983461856841, + "grad_norm": 0.0004980686353519559, + "kept_token_count": 572.5640625, + "learning_rate": 2.758e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9473459899425507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00020567993924487383, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.4051383972168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9757843017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1380, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2991992324590683, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.10942912697792054, + "grad_norm": 0.0006416601245291531, + "kept_token_count": 573.3453125, + "learning_rate": 2.778e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9475864410400391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00020474065822781995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.50436935424804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.08730163574216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1390, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29990905821323394, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.1073506385087967, + "grad_norm": 0.0004887873074039817, + "kept_token_count": 573.0203125, + "learning_rate": 2.798e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9470053374767303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00020701060711871832, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.09496536254883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1028717041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1400, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.30028378069400785, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.10578953251242637, + "grad_norm": 0.000582357868552208, + "kept_token_count": 572.6046875, + "learning_rate": 2.818e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9496606767177582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019663822458824143, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.10496673583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.82341918945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1410, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.300701892375946, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.11929982230067253, + "grad_norm": 0.0006207146216183901, + "kept_token_count": 572.146875, + "learning_rate": 2.838e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9492410778999328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019827724754577502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.56575088500976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7386474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1420, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3001101642847061, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.12418899312615395, + "grad_norm": 0.0007568129221908748, + "kept_token_count": 572.6234375, + "learning_rate": 2.858e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9480795800685883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00020281429606257008, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.80003356933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.85442810058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3003893464803696, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.11566836461424827, + "grad_norm": 0.0006652325973846018, + "kept_token_count": 572.4, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9507317125797272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019245443399995565, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.6486442565918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5561828613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1440, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.29942484200000763, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.12002184689044952, + "grad_norm": 0.0005057601374574006, + "kept_token_count": 573.3015625, + "learning_rate": 2.898e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9527711749076844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001844877187977545, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 101.79710311889649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.66631774902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1450, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3000250428915024, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.1289985477924347, + "grad_norm": 0.0006121451151557267, + "kept_token_count": 572.6484375, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9505182683467865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019328822090756149, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.45704803466796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4388488769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1460, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.3007245004177094, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.11770683750510216, + "grad_norm": 0.0007865314255468547, + "kept_token_count": 572.228125, + "learning_rate": 2.938e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9524620711803437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00018569518579170108, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.74534454345704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4384338378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1470, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30015183091163633, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.12025767490267754, + "grad_norm": 0.0005784580134786665, + "kept_token_count": 572.7984375, + "learning_rate": 2.958e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.952025544643402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00018740041414275766, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.08899765014648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68515930175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1480, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2998544484376907, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.11650893539190292, + "grad_norm": 0.0007396719884127378, + "kept_token_count": 572.875, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.952943605184555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001838142517954111, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 101.295751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3000061035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1490, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.29985029697418214, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.12839823737740516, + "grad_norm": 0.0007525277324020863, + "kept_token_count": 573.0421875, + "learning_rate": 2.998e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9512963116168975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019024891662411392, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.12759246826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6251678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1500, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.300691682100296, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.12062831521034241, + "grad_norm": 0.000500797585118562, + "kept_token_count": 571.903125, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9512414515018464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00019046321540372446, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 102.05396423339843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.01067504882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1510, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.2992689907550812, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.11553823873400688, + "grad_norm": 0.0005488378810696304, + "kept_token_count": 573.309375, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9515946686267853, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00018908352649305016, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.51565017700196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.48108825683596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1520, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3008082687854767, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.11452677696943284, + "grad_norm": 0.0007057291804812849, + "kept_token_count": 572.296875, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9560461938381195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00017169469210784883, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.63580017089843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.23264770507814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1530, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2994110405445099, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.12726491317152977, + "grad_norm": 0.0007169898599386215, + "kept_token_count": 573.31875, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9521829605102539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001867855593445711, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.42872009277343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2704315185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1540, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.30026120245456694, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.12581735402345656, + "grad_norm": 0.0005512605421245098, + "kept_token_count": 572.8984375, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9578116834163666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016479834594065323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 108.2878646850586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.05431213378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1550, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.3001042455434799, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.12683942019939423, + "grad_norm": 0.0006737990188412368, + "kept_token_count": 572.5609375, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9578001022338867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016484358347952367, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 108.07223587036133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0030883789062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1560, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.29888429641723635, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.1282837860286236, + "grad_norm": 0.0005790515569970012, + "kept_token_count": 573.753125, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9555928409099579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00017346571694361047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 107.62813796997071, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8060668945312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1570, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30013747215270997, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.1332128643989563, + "grad_norm": 0.0006668822024948895, + "kept_token_count": 572.5203125, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9563720345497131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00017042189283529298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.49163055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2514709472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1580, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.2998283296823502, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.1264921322464943, + "grad_norm": 0.000686424202285707, + "kept_token_count": 573.078125, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9563929915428162, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00017034007469192148, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.45300750732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.02068786621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1590, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.2995101034641266, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.13253097906708716, + "grad_norm": 0.0004896594327874482, + "kept_token_count": 573.25625, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.957829874753952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016472723364131526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 112.96925659179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7778015136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1600, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.299967485666275, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.12165199220180511, + "grad_norm": 0.000560408050660044, + "kept_token_count": 572.80625, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.957705271244049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016521396464668213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.11126251220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8478973388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1610, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29941627085208894, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.11968781873583793, + "grad_norm": 0.0005224290071055293, + "kept_token_count": 573.0125, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9576166093349456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001655602507526055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 114.16059494018555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31529541015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1620, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.2993420034646988, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.12182342335581779, + "grad_norm": 0.0004501747025642544, + "kept_token_count": 573.178125, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9583521842956543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016268695326289163, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 113.69736022949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.71543273925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3000867873430252, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.11748139634728431, + "grad_norm": 0.0006563295610249043, + "kept_token_count": 572.78125, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9591058135032654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015974311827449128, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 113.65451278686524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.75400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.299000671505928, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.11945428028702736, + "grad_norm": 0.0006053695688024163, + "kept_token_count": 573.5203125, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9579185426235199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016438087914139032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.92663269042968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0628601074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2999163746833801, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.12443682178854942, + "grad_norm": 0.0004830078105442226, + "kept_token_count": 572.834375, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9594692707061767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015832338976906613, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.91773681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21564331054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1660, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2985441625118256, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.12804915606975556, + "grad_norm": 0.0005957140238024294, + "kept_token_count": 574.2265625, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9579830884933471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000164128742471803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.66643981933593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5766937255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1670, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.3000509262084961, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.11760296002030372, + "grad_norm": 0.0006210977444425225, + "kept_token_count": 572.796875, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9597082197666168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015738997026346624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 120.99336318969726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9822174072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1680, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.29952468276023864, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.13540633991360665, + "grad_norm": 0.0004704361199401319, + "kept_token_count": 573.078125, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9586954474449157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00016134608595166355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.22833099365235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2136962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2985377967357635, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.13355439230799676, + "grad_norm": 0.00047757025458849967, + "kept_token_count": 574.021875, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9601663291454315, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000155600490688812, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 117.26424865722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4514129638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2995794892311096, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.12493072003126145, + "grad_norm": 0.0004320473817642778, + "kept_token_count": 573.1109375, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9593254685401916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001588850383996032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 117.83405151367188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55699157714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1710, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3002367913722992, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.12934158593416215, + "grad_norm": 0.0006203013472259045, + "kept_token_count": 572.4859375, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9630731105804443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014424581968341954, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 121.30972061157226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.64427185058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1720, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29929311871528624, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.13116029128432274, + "grad_norm": 0.0004686246393248439, + "kept_token_count": 573.2921875, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9594557106494903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001583763267262839, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 120.0289794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1355926513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1730, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2996119886636734, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.11846031099557877, + "grad_norm": 0.0005355190369300544, + "kept_token_count": 573.11875, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9606266498565674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015380235126940533, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 119.17176742553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.29795227050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1740, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.300589320063591, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.1342276707291603, + "grad_norm": 0.0005780716892331839, + "kept_token_count": 572.315625, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9592337071895599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015924361650831997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.10141220092774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24925537109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1750, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2998260796070099, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.116644848883152, + "grad_norm": 0.00041299505392089486, + "kept_token_count": 573.0578125, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9634242057800293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014287435624282807, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.0876838684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4143463134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1760, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.2984337449073792, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.12891209721565247, + "grad_norm": 0.00046378106344491243, + "kept_token_count": 573.84375, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9599724113941193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015635792806278914, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.22812957763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.07899475097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1770, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.3003253012895584, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.12954872772097586, + "grad_norm": 0.0005019225645810366, + "kept_token_count": 572.5484375, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9626718163490295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014581341820303352, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.76127471923829, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8582336425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1780, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.29933765828609465, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.13083014786243438, + "grad_norm": 0.0005426137940958142, + "kept_token_count": 573.259375, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9627020418643951, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001456952581065707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.75983123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7236663818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1790, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2994126081466675, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.12980877831578255, + "grad_norm": 0.0005148020572960377, + "kept_token_count": 573.2734375, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9617980897426606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014922642440069468, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.8886619567871, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6883056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1800, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30095346868038175, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.1196078784763813, + "grad_norm": 0.0005988148041069508, + "kept_token_count": 571.940625, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9628332793712616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014518271200358868, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.28479766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4623779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1810, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2996691793203354, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.12809145003557204, + "grad_norm": 0.00036816776264458895, + "kept_token_count": 573.0734375, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9599324047565461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015651425055693836, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.57515182495118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8277099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1820, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2996564120054245, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.11295583918690681, + "grad_norm": 0.0005118520930409431, + "kept_token_count": 573.325, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9622579395771027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014743006322532893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.16599807739257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.05292053222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1830, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.3015535116195679, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.1259017638862133, + "grad_norm": 0.000832865247502923, + "kept_token_count": 571.5875, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9610062420368195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001523195460322313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 127.94973907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1499328613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1840, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2995665192604065, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.13655710965394974, + "grad_norm": 0.0005247078370302916, + "kept_token_count": 572.9890625, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9605275392532349, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015418951225001364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 133.29269256591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.56488037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1850, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.300616979598999, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.12115626782178879, + "grad_norm": 0.000549874035641551, + "kept_token_count": 572.1359375, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0002, + "mgd_cos_sim": 0.9609214007854462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00015265096662915313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 127.27064361572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.75543518066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1860, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.2996131509542465, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.11968232095241546, + "grad_norm": 0.0004235405649524182, + "kept_token_count": 573.1390625, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9641537427902221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001400246263074223, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 131.80235214233397, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0843505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1870, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.3002506345510483, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.13010523840785027, + "grad_norm": 0.0005076905945315957, + "kept_token_count": 572.6953125, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9643753170967102, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013915917006670497, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.2594955444336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.12407836914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1880, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3005773812532425, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.13385010957717897, + "grad_norm": 0.0005062196287326515, + "kept_token_count": 572.475, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9623287141323089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014715365978190675, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 128.5332046508789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.59056701660154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1890, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.2997177571058273, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.12328973412513733, + "grad_norm": 0.0004599677340593189, + "kept_token_count": 573.0890625, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9625642657279968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001462335276301019, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 127.53162231445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4103210449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1900, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.300256559252739, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.127826476842165, + "grad_norm": 0.0004659533442463726, + "kept_token_count": 572.59375, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9640444338321685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001404516107868403, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 136.16583557128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4396545410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2993209809064865, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.12805367931723594, + "grad_norm": 0.0005658618174493313, + "kept_token_count": 573.2171875, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9616363763809204, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014985814195824786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 131.7190170288086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4049774169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3004273533821106, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.12527616024017335, + "grad_norm": 0.000521595124155283, + "kept_token_count": 572.1984375, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9626360893249511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014595296670449898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 130.75749664306642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1846923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1930, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.30004113018512724, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.12373376935720444, + "grad_norm": 0.0005201941821724176, + "kept_token_count": 572.7640625, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9645693480968476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013840119063388557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.28059463500978, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6713470458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30075311064720156, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.12514434307813643, + "grad_norm": 0.0004378652083687484, + "kept_token_count": 572.165625, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9660339653491974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013268002512631938, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.02826080322265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2375091552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1950, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.30080808997154235, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.1250057838857174, + "grad_norm": 0.000447993166744709, + "kept_token_count": 572.18125, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9680492341518402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012480786317610183, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 133.70147705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.43236083984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1960, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3009751349687576, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.1367374800145626, + "grad_norm": 0.0006387775647453964, + "kept_token_count": 572.03125, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9653386890888214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001353959050902631, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 136.09833526611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.612255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1970, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.29973390996456145, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.1332710824906826, + "grad_norm": 0.0004555082123260945, + "kept_token_count": 573.1703125, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9627207100391388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001456224694265984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 131.95186767578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1162353515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1980, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.3001027911901474, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.12301161661744117, + "grad_norm": 0.0005312206922098994, + "kept_token_count": 572.7328125, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9625375628471374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014633777900598943, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 130.6102325439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8191162109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1990, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.299291130900383, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.11858453527092934, + "grad_norm": 0.0004909996641799808, + "kept_token_count": 573.365625, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9659436345100403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013303294108482077, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.3357711791992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2438598632813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2000, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2996525287628174, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.13570669516921044, + "grad_norm": 0.0004966175765730441, + "kept_token_count": 572.9125, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9627877593040466, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014536049711750821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.92745361328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.02931823730466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.300545534491539, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.12521629631519318, + "grad_norm": 0.00048616345156915486, + "kept_token_count": 572.2828125, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9660582900047302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013258501567179338, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.16879425048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.36571350097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2020, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3007037669420242, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.13426267132163047, + "grad_norm": 0.0006365119479596615, + "kept_token_count": 572.115625, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9647617399692535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013764965115115047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 138.75302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.24518127441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2030, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3001347243785858, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.12430412694811821, + "grad_norm": 0.00045619060983881354, + "kept_token_count": 572.5984375, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9657142758369446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013392885884968563, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.02108459472657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0976043701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2040, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3006300121545792, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.13429844677448272, + "grad_norm": 0.000612205418292433, + "kept_token_count": 572.25, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9646497905254364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013808690709993243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.9165496826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.40663146972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3000397354364395, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.12129052355885506, + "grad_norm": 0.00044974498450756073, + "kept_token_count": 572.7546875, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9682201743125916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012414016018738038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.89344787597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.57944641113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2060, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30021525323390963, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.14078519567847253, + "grad_norm": 0.0004721459117718041, + "kept_token_count": 572.496875, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9617968916893005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014923101844033226, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.33856811523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.50791320800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2070, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.3004857063293457, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.12070507109165192, + "grad_norm": 0.0004920241190120578, + "kept_token_count": 572.4359375, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9642857074737549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013950912398286164, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.40052795410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.84083862304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2080, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2997561186552048, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.12806807905435563, + "grad_norm": 0.0004824507632292807, + "kept_token_count": 572.7953125, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9661580860614777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013219518805271946, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.85913543701173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4395812988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2090, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29889340698719025, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.12824157625436783, + "grad_norm": 0.00037742318818345666, + "kept_token_count": 573.828125, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9654862821102143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013481945352395998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.50725708007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9702087402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2100, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.29939842224121094, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.12224340438842773, + "grad_norm": 0.0005613519460894167, + "kept_token_count": 573.3328125, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.964757764339447, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013766523479716853, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.68301391601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21995239257814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2110, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30022635161876676, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.12749461084604263, + "grad_norm": 0.000427661114372313, + "kept_token_count": 572.671875, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9680892825126648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012465140971471556, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.96088409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0307586669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2120, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.30023022890090945, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.12408940494060516, + "grad_norm": 0.0004763927427120507, + "kept_token_count": 572.7359375, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9636327683925628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00014205968182068318, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 141.06920623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7589050292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2130, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.29947046339511874, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.1279062256217003, + "grad_norm": 0.0004741602169815451, + "kept_token_count": 573.090625, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9667421877384186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012991351104574278, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.6273681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7294067382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.30052198469638824, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.12313791438937187, + "grad_norm": 0.0005712839192710817, + "kept_token_count": 572.4859375, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9680000126361847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012500011071097105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.1132598876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.04356079101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2150, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.30014853477478026, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.11915606632828712, + "grad_norm": 0.0004392618720885366, + "kept_token_count": 572.740625, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9680957436561585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012462618324207141, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.50433959960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7466674804688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2160, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.3004522293806076, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.11668552532792091, + "grad_norm": 0.00046529978862963617, + "kept_token_count": 572.309375, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9683617293834687, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012358715102891437, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.84066925048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4062438964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2170, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3004747271537781, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.1302293933928013, + "grad_norm": 0.0005662919138558209, + "kept_token_count": 572.4703125, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9658622026443482, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013335097464732825, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.61156616210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2180, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.30031531751155854, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.12202555835247039, + "grad_norm": 0.00047673325752839446, + "kept_token_count": 572.3859375, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9667159140110015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013001610932406038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.41759033203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01019897460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3004321277141571, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.13838267251849173, + "grad_norm": 0.00039565438055433333, + "kept_token_count": 572.61875, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9670480489730835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000128718807536643, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.0489944458008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6050170898437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2200, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30054489374160764, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.1275734469294548, + "grad_norm": 0.0003899075381923467, + "kept_token_count": 572.565625, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9682660102844238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012396105739753693, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.53093566894532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4722137451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2210, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.3000478744506836, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.12720355167984962, + "grad_norm": 0.0005081294220872223, + "kept_token_count": 572.875, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9686633944511414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012240879514138214, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.00680389404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.72291870117186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2220, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.29969159364700315, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.1295579344034195, + "grad_norm": 0.0005438756197690964, + "kept_token_count": 573.203125, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9660479068756104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001326255252934061, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.32381439208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.98174743652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2230, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.298915958404541, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.1344602718949318, + "grad_norm": 0.00043113803258165717, + "kept_token_count": 573.8328125, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9676700174808502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012628915210370905, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.9417709350586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6127105712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2240, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.2993896633386612, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.12604195550084113, + "grad_norm": 0.000517335080076009, + "kept_token_count": 573.2875, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9663187325000763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00013156759305275046, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 141.14295196533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55421752929686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3001043438911438, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.12551581785082816, + "grad_norm": 0.00037400564178824425, + "kept_token_count": 572.6796875, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9687608182430267, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012202830257592722, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.1827133178711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7764007568359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2260, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.29993914663791654, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.1171131506562233, + "grad_norm": 0.00046364302397705615, + "kept_token_count": 572.7109375, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9682437241077423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012404814769979567, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.25733337402343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6771514892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2270, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2995624035596848, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.12035142332315445, + "grad_norm": 0.00037152605364099145, + "kept_token_count": 573.134375, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9678539574146271, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001255706614756491, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.96689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8826568603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2280, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30011430382728577, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.12648279294371606, + "grad_norm": 0.0004179663665127009, + "kept_token_count": 572.6484375, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9679583311080933, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012516299102571793, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.2811264038086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2061492919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30084988474845886, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.13206401094794273, + "grad_norm": 0.000500058988109231, + "kept_token_count": 571.8609375, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9676437318325043, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001263918136828579, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.37578887939452, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6718322753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2300, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.30058022141456603, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.13170647770166397, + "grad_norm": 0.00044337857980281115, + "kept_token_count": 572.53125, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9678556144237518, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012556420479086227, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.41563262939454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.11610412597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2310, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.29943702518939974, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.1254543162882328, + "grad_norm": 0.00039501162245869637, + "kept_token_count": 573.234375, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9689821898937225, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012116348152630962, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.4381576538086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3093231201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2993656754493713, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.12849830389022826, + "grad_norm": 0.0005398663925006986, + "kept_token_count": 573.2546875, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9680802822113037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012468655040720478, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.34971771240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0054168701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2330, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.29975171089172364, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.12928904891014098, + "grad_norm": 0.0005637174472212791, + "kept_token_count": 573.1515625, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9681138515472412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001245554376509972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.55062561035157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.27002868652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2992866367101669, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.13550099804997445, + "grad_norm": 0.0005077740643173456, + "kept_token_count": 573.2171875, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9684422791004181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012327253498369827, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.04811401367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3851379394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2350, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.2999796748161316, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.13187981247901917, + "grad_norm": 0.0004997918149456382, + "kept_token_count": 572.7875, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9679574310779572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000125166516227182, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.67823944091796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0412170410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2993677169084549, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.1336019217967987, + "grad_norm": 0.00040335801895707846, + "kept_token_count": 573.2921875, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9678100109100342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012574227439472453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.99541625976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5750030517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2370, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2990987241268158, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.12317979782819748, + "grad_norm": 0.0003026121703442186, + "kept_token_count": 573.6140625, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9707223057746888, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001143661305832211, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.09605255126954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6454742431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2380, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3010799288749695, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.1340843752026558, + "grad_norm": 0.0004954456817358732, + "kept_token_count": 572.203125, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9708355069160461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011392399246688001, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.1553176879883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.07237548828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2390, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.30087811350822447, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.12821680754423143, + "grad_norm": 0.00038606490124948323, + "kept_token_count": 572.2765625, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9704390048980713, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011547288740985096, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.3958312988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.99033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2400, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.30113864839077, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.124187533557415, + "grad_norm": 0.0004289760545361787, + "kept_token_count": 571.8859375, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9714865028858185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011138102054246702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.99942474365236, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.88262939453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2410, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.300023353099823, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.13477245569229127, + "grad_norm": 0.00043522592750377953, + "kept_token_count": 572.6265625, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9687430560588837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012209762790007516, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.41123657226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4697204589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2420, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.29945936501026155, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.12671801596879959, + "grad_norm": 0.0005182691384106874, + "kept_token_count": 573.046875, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9694704949855805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011925607832381502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.40933990478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7566223144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2430, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.2996580570936203, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.12130216658115386, + "grad_norm": 0.00044005856034345925, + "kept_token_count": 573.2125, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9700728535652161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011690308747347444, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.3025161743164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3365997314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2440, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.3004885256290436, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.13751318231225013, + "grad_norm": 0.00044819628237746656, + "kept_token_count": 572.5984375, + "learning_rate": 2.991800117455e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9699471652507782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.000117394018889172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.82219543457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7764892578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2450, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.2999848395586014, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.13208834752440451, + "grad_norm": 0.0004196477239020169, + "kept_token_count": 573.1546875, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9702248334884643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011630942171905189, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.0171920776367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3144287109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2460, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.29960799515247344, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.13668132498860358, + "grad_norm": 0.0004754411638714373, + "kept_token_count": 573.4734375, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9698102951049805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011792875593528152, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.32150268554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3933898925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2470, + "valid_token_count": 818.7953125 + }, + { + "actual_token_mask_ratio": 0.29937220811843873, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.12696603536605836, + "grad_norm": 0.00037241014069877565, + "kept_token_count": 573.328125, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9704261481761932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011552303767530248, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.33991241455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.47214965820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2480, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3009701281785965, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.12550228461623192, + "grad_norm": 0.00044792951666750014, + "kept_token_count": 572.128125, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9723047256469727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010818478840519674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.60128479003907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2473449707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2490, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2999018520116806, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.12877496629953383, + "grad_norm": 0.0005533821531571448, + "kept_token_count": 572.8890625, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.968437272310257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012329209203016945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.94493408203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3233306884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2500, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29976661801338195, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.13329235911369325, + "grad_norm": 0.0004274066013749689, + "kept_token_count": 573.153125, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9692164123058319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012024856259813532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.1645706176758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75184020996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2510, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2992947638034821, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.1335909768939018, + "grad_norm": 0.00043867406202480197, + "kept_token_count": 573.296875, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9709413290023804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011351062858011574, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.2864761352539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3577453613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2991831302642822, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.1344915173947811, + "grad_norm": 0.00048769405111670494, + "kept_token_count": 573.5359375, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9714835405349731, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011139258349430747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.20366973876952, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.17425537109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2530, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.30020748674869535, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.12273642644286156, + "grad_norm": 0.0004756119742523879, + "kept_token_count": 572.4921875, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9707808017730712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011413770334911533, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.1734146118164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8063629150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2540, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2988709002733231, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.14025365337729453, + "grad_norm": 0.00045218117884360254, + "kept_token_count": 573.4828125, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9690986454486847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012070860175299458, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.9244598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2394561767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2550, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.300030916929245, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.13094521015882493, + "grad_norm": 0.0004357992729637772, + "kept_token_count": 572.775, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9692591845989227, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00012008154881186784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.9860397338867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.175048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2560, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3008712649345398, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.12767477631568908, + "grad_norm": 0.0004414902941789478, + "kept_token_count": 572.2140625, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9711591601371765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011265972352703102, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.74080505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.10621948242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2570, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.299029803276062, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.13221280500292779, + "grad_norm": 0.00043265309068374336, + "kept_token_count": 573.7765625, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9714122593402863, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011167097400175408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.42679290771486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3106994628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2580, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.30001520216464994, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.11948327571153641, + "grad_norm": 0.0004132043104618788, + "kept_token_count": 572.7140625, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9706077992916107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011481345864012837, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.8732437133789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.35328063964846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2590, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.299760702252388, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.1373061716556549, + "grad_norm": 0.00038984703132882714, + "kept_token_count": 573.0203125, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9711995422840118, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011250197057961486, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.08354797363282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7353485107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2600, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2999037325382233, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.12691569775342942, + "grad_norm": 0.0003844194579869509, + "kept_token_count": 572.821875, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9721344828605651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010884985895245335, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.23669891357423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3882659912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2610, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2998520076274872, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.12173645794391633, + "grad_norm": 0.0005029381718486547, + "kept_token_count": 572.7015625, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9710014879703521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011327558750053867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.08052825927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.74928283691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2620, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.29940208196640017, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.1219266802072525, + "grad_norm": 0.00038281569140963256, + "kept_token_count": 573.23125, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9724391639232636, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010765964470920153, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.05348510742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8110412597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2630, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3013270109891891, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.1304153323173523, + "grad_norm": 0.0004848993557970971, + "kept_token_count": 571.9421875, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9710877716541291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011293865682091564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.79273834228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1228729248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2640, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.2992942839860916, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.12955179959535598, + "grad_norm": 0.0006339699029922485, + "kept_token_count": 573.2875, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9717988669872284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011016082280548289, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.24497985839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.273388671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3004018247127533, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.1322772592306137, + "grad_norm": 0.00042597821447998285, + "kept_token_count": 572.3859375, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9697279393672943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011825040855910629, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.60357818603515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.81812744140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2660, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30010572969913485, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.12967915311455727, + "grad_norm": 0.0004903690423816442, + "kept_token_count": 572.528125, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9697436153888702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011818918937933631, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.03076629638673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.73212280273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2670, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.30002304911613464, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.12785936146974564, + "grad_norm": 0.0004769926890730858, + "kept_token_count": 572.740625, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9723331451416015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010807380749611184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.6346435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7341247558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2680, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3002796441316605, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.12960849180817605, + "grad_norm": 0.00035068936995230615, + "kept_token_count": 572.715625, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9727035582065582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010662689746823161, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.2720504760742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.879345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2690, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.2999670207500458, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.14278150498867034, + "grad_norm": 0.00045639750896953046, + "kept_token_count": 572.6171875, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9715722262859344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011104617224191315, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.92083740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17881469726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2700, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.300745290517807, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.13771385923027993, + "grad_norm": 0.0004530660808086395, + "kept_token_count": 572.2609375, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9720967411994934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010899729313678108, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.2403533935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.70546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2710, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2991242378950119, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.13328757733106614, + "grad_norm": 0.00035311139072291553, + "kept_token_count": 573.3765625, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9710248649120331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011318431206746027, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.02307586669923, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3301635742188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.29982468783855437, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.13237039744853973, + "grad_norm": 0.0006005433388054371, + "kept_token_count": 572.9140625, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9712755978107452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001122049019613769, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.70826263427733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.92449951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2730, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29945768117904664, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.1263076938688755, + "grad_norm": 0.0004284144670236856, + "kept_token_count": 573.021875, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9730831027030945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010514434252399951, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.12962036132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4779907226563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2740, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.2992925137281418, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.12103672176599503, + "grad_norm": 0.0003800983540713787, + "kept_token_count": 573.2328125, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9723166525363922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010813829358085059, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.5015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70490112304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2750, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.3001953333616257, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.12436997741460801, + "grad_norm": 0.0003813967341557145, + "kept_token_count": 572.7125, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9717025995254517, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011053685011574998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.71035766601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6707824707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2760, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2994206815958023, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.12726899236440659, + "grad_norm": 0.0004564962873701006, + "kept_token_count": 573.0046875, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9728741466999054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010596057691145689, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.6255676269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8066925048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2770, + "valid_token_count": 817.9015625 + }, + { + "actual_token_mask_ratio": 0.2992840319871902, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.13031734973192216, + "grad_norm": 0.0004631958727259189, + "kept_token_count": 573.059375, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9725911378860473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010706608882173895, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.84894561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.02427368164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2780, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.29940307438373565, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.12526417374610901, + "grad_norm": 0.0004070932627655566, + "kept_token_count": 573.171875, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9714718878269195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011143810479552485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 166.78728790283202, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9494659423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.29984906911849973, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.11907594949007035, + "grad_norm": 0.0004160107346251607, + "kept_token_count": 572.6875, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9741340517997742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010103907843586057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.81862182617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.8524444580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2800, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.30001914203166963, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.14179206937551497, + "grad_norm": 0.0005520195118151605, + "kept_token_count": 572.9125, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9710427224636078, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011311454654787667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.89203948974608, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0990844726563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2810, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.30005823373794555, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.13133564740419387, + "grad_norm": 0.0004704846360255033, + "kept_token_count": 572.8015625, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9713120043277741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011206265480723232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.9144317626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9390808105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2820, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2981425732374191, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.12956997230648995, + "grad_norm": 0.0004743437166325748, + "kept_token_count": 574.4828125, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.97412810921669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010106226982316002, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 166.66846160888673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8596923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2830, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.2998128294944763, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.12515943497419357, + "grad_norm": 0.0004417596210259944, + "kept_token_count": 572.9671875, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9729553997516632, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010564316689851694, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.65436706542968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.03014221191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2840, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.30046064853668214, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.12759524211287498, + "grad_norm": 0.0003783504944294691, + "kept_token_count": 572.4546875, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9739994525909423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010156479911529459, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 166.22147369384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82017211914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2850, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30035450160503385, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.13809083998203278, + "grad_norm": 0.0003627588739618659, + "kept_token_count": 572.6671875, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9742443561553955, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010060817439807578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.95848846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1154510498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2860, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.30037143230438235, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.13115817829966545, + "grad_norm": 0.00032544051646254957, + "kept_token_count": 572.3375, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9742777824401856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010047762043541297, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.66522064208985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8303588867187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2870, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.3002631664276123, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.12664921209216118, + "grad_norm": 0.0003798134857788682, + "kept_token_count": 572.4703125, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9721296310424805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010886878808378242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.1867706298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.23138122558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2880, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.2991547018289566, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.13693944364786148, + "grad_norm": 0.0004290487850084901, + "kept_token_count": 573.4796875, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9709099590778351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011363318699295633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.24904022216796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2245300292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2890, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.299508199095726, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.12176102325320244, + "grad_norm": 0.0005537058459594846, + "kept_token_count": 573.0, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9736410796642303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010296466498402879, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.1373779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.67088928222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2900, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.3002481788396835, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.12621424719691277, + "grad_norm": 0.000564045156352222, + "kept_token_count": 572.753125, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.973131287097931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010495605529285968, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.76776428222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7198791503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2910, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.29966410994529724, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.13848934322595596, + "grad_norm": 0.000426966609666124, + "kept_token_count": 572.9375, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9715153396129608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011126837198389694, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.06015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.73650512695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.299024373292923, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.12611539512872696, + "grad_norm": 0.0004960038932040334, + "kept_token_count": 573.8359375, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9723167717456818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010813780900207349, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.69568634033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34364624023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2930, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.30053292512893676, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.1347704492509365, + "grad_norm": 0.0005455760401673615, + "kept_token_count": 572.4078125, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9739397406578064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010179811652051285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.84155731201173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2940, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2991710722446442, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.12875425964593887, + "grad_norm": 0.0005188444047234952, + "kept_token_count": 573.684375, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9738657832145691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010208697858615778, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.83641967773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.39820556640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2950, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.2998619318008423, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.13028107285499574, + "grad_norm": 0.0003655898035503924, + "kept_token_count": 573.028125, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.974252051115036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010057814652100205, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.5109832763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2650482177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2960, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.2998824089765549, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.13439000844955445, + "grad_norm": 0.0004984423285350204, + "kept_token_count": 572.8875, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9736370265483856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001029805040161591, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.50256805419923, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7530090332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30068196952342985, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.1390514299273491, + "grad_norm": 0.00039835876668803394, + "kept_token_count": 572.33125, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9726301014423371, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010691388742998243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.956591796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.82745361328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.30034945011138914, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.12047386541962624, + "grad_norm": 0.00034709679312072694, + "kept_token_count": 572.3140625, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9733239412307739, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010420359976706094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.26539154052733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.13819274902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2990, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.29985139071941375, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.11690196692943573, + "grad_norm": 0.0004359539598226547, + "kept_token_count": 573.2015625, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9751699566841125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.699259026092477e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.19109191894532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.34343872070315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3000, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.2999149411916733, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.12888167351484298, + "grad_norm": 0.00042538123670965433, + "kept_token_count": 572.9421875, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9738964796066284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010196703224210068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.30624389648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77867431640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3010, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2984332174062729, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.12626441493630408, + "grad_norm": 0.0003892872482538223, + "kept_token_count": 574.40625, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9724338471889495, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010768045685836115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.21002349853515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.40076599121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3020, + "valid_token_count": 818.7578125 + }, + { + "actual_token_mask_ratio": 0.3005707114934921, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.12460141479969025, + "grad_norm": 0.00031806298648007214, + "kept_token_count": 572.4828125, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9743287622928619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010027846437878906, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.49669342041017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.53167114257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3030, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2990740239620209, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.13518959283828735, + "grad_norm": 0.0003717124927788973, + "kept_token_count": 573.69375, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9733313322067261, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010417473240522667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.21588439941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3155456542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3040, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2996397465467453, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.12369277626276017, + "grad_norm": 0.0005138724809512496, + "kept_token_count": 572.853125, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9735171556472778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010344879265176133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.09059143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8958984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3050, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.3007557839155197, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.13173319548368453, + "grad_norm": 0.000527195748873055, + "kept_token_count": 571.9171875, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9718098282814026, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00011011809619958513, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.87950744628907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9049438476562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3060, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.30086635053157806, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.12454351112246513, + "grad_norm": 0.00042386239510960877, + "kept_token_count": 572.0828125, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9758119344711303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.448485579923726e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.15813293457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.16382751464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3070, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2995898574590683, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.12849077954888344, + "grad_norm": 0.00036884646397084, + "kept_token_count": 572.9265625, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9742994904518127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010039277258329093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.01732025146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0884063720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30066893696784974, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.13498812690377235, + "grad_norm": 0.0005025299033150077, + "kept_token_count": 572.1078125, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9763563930988312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.235803954652511e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.25738983154298, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.45787353515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3090, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.29931350946426394, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.1437722772359848, + "grad_norm": 0.00046966419904492795, + "kept_token_count": 573.1765625, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9740990698337555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001011756554362364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.3519302368164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5362243652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3100, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.30009226202964784, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.13707346990704536, + "grad_norm": 0.00043693120824173093, + "kept_token_count": 572.775, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9742037892341614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010076665785163641, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.23422088623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7812255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3110, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30011603236198425, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.1352284200489521, + "grad_norm": 0.0003367952012922615, + "kept_token_count": 572.8, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.975161737203598, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.702472452772781e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.5864501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5672302246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3120, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2991386502981186, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.13695584684610368, + "grad_norm": 0.000446065969299525, + "kept_token_count": 573.4015625, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9749175488948822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.79784905212e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.03857879638673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4988098144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3130, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.3003746956586838, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.13583453968167306, + "grad_norm": 0.00045077758841216564, + "kept_token_count": 572.48125, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9749108493328095, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.800474072108045e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.2602081298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1285766601562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3140, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29943830966949464, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.12831236720085143, + "grad_norm": 0.0004681205900851637, + "kept_token_count": 573.325, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9746636509895324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.897025884129108e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.35748748779298, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8252014160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3150, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.29911606311798095, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.12988248765468596, + "grad_norm": 0.0004981863894499838, + "kept_token_count": 573.621875, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9744858980178833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.966463112505153e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.56531219482423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.37840576171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3160, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2995746821165085, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.12797385081648827, + "grad_norm": 0.0004057154292240739, + "kept_token_count": 573.1390625, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9763636767864228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.23295781831257e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.04439544677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4626220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3170, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2998835355043411, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.128596693277359, + "grad_norm": 0.0004298020212445408, + "kept_token_count": 572.9765625, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9748375833034515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.829084883676842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.3472686767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.69630737304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3180, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29968800842761995, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.1293290674686432, + "grad_norm": 0.0005008145235478878, + "kept_token_count": 572.9859375, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9736150622367858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0001030662955599837, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.25543975830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.32050170898435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3190, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.29889683425426483, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.12891051024198533, + "grad_norm": 0.00044855382293462753, + "kept_token_count": 573.7921875, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9740652441978455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010130784503417089, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.86515197753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.3084320068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3200, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2999852508306503, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.13696220889687538, + "grad_norm": 0.0004718090931419283, + "kept_token_count": 572.778125, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9738145530223846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010228715691482649, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.66551513671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.94532775878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29993622601032255, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.128825993090868, + "grad_norm": 0.0004107590648345649, + "kept_token_count": 572.775, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9754529237747193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.588722386979497e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.1171905517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.31732177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3220, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.30032884180545805, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.12443822249770164, + "grad_norm": 0.0004898747429251671, + "kept_token_count": 572.4875, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9769572854042053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.001081343740225e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.56776123046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0026428222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3230, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30077198445796965, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.12373526096343994, + "grad_norm": 0.0004830880498047918, + "kept_token_count": 571.9078125, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9772906124591827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.870877791196108e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.76306762695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.29799499511716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3240, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.2998925060033798, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.13232773169875145, + "grad_norm": 0.0004125010163988918, + "kept_token_count": 572.94375, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9739467918872833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010177058575209231, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.93624725341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.49815979003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3250, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29934745132923124, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.13664290755987168, + "grad_norm": 0.00038101867539808154, + "kept_token_count": 573.2859375, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9753418862819672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.632098881411366e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.17149200439454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8633209228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3260, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3005348712205887, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.12919967696070672, + "grad_norm": 0.0005688229575753212, + "kept_token_count": 572.5625, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9756387531757355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.516131322016009e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.09677276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.15082397460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3270, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30026088654994965, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.14391666129231453, + "grad_norm": 0.000457358721178025, + "kept_token_count": 572.503125, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9741831362247467, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00010084730383823626, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.90911865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.41502380371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3280, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30138435065746305, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.14193989485502242, + "grad_norm": 0.00042268758988939226, + "kept_token_count": 571.684375, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.97451953291893, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.953327898983843e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.3231216430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8933959960938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3290, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.30006093382835386, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.1289970502257347, + "grad_norm": 0.0003470519732218236, + "kept_token_count": 572.8390625, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9758637726306916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.428228950127959e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.9974578857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.23882751464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3300, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2999785304069519, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.13093543723225592, + "grad_norm": 0.00037862895987927914, + "kept_token_count": 572.778125, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9772115170955658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.901767869247123e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.89640350341796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9991790771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3310, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2994576901197433, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.1359122261404991, + "grad_norm": 0.000391041103284806, + "kept_token_count": 573.3046875, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9761381089687348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.321073666797019e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.69259643554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7138305664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3320, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.3001790016889572, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.13439571037888526, + "grad_norm": 0.0004052543663419783, + "kept_token_count": 572.59375, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9754958808422088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.571938644512557e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.26516876220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5918365478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3330, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29989110231399535, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.1259033888578415, + "grad_norm": 0.0003697905340231955, + "kept_token_count": 572.6625, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9761346220970154, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.322437617811374e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.44164581298827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1234130859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3340, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.29995750784873965, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.13522203117609025, + "grad_norm": 0.0003566164232324809, + "kept_token_count": 572.665625, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9761509537696839, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.316047871834598e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.3761993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.19227905273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3350, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.2987625032663345, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.1278445802628994, + "grad_norm": 0.0004888048861175776, + "kept_token_count": 573.6421875, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9759712815284729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.386236342834309e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.5545211791992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.73688049316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3360, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.2999908208847046, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.13792526200413704, + "grad_norm": 0.00034303293796256185, + "kept_token_count": 572.7640625, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9755993545055389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.531521354801952e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.58682098388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.974755859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3370, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3003097474575043, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.12761607244610787, + "grad_norm": 0.0003533865383360535, + "kept_token_count": 572.621875, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9767939865589141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.064870901056565e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.12349700927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0024047851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3380, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30052622854709626, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.12320708855986595, + "grad_norm": 0.000359470141120255, + "kept_token_count": 572.2609375, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9760809302330017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.34340809180867e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.5804214477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8988800048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3390, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2987008810043335, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.14129195138812065, + "grad_norm": 0.00041604082798585296, + "kept_token_count": 573.6015625, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9760746538639069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.345864527858794e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.36056671142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.17498779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3400, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.3005653828382492, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.1240464374423027, + "grad_norm": 0.00039652406121604145, + "kept_token_count": 572.2265625, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9767304480075836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.089689629036002e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.9192123413086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.26134033203124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3410, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2990131855010986, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.1122740387916565, + "grad_norm": 0.00037154488381929696, + "kept_token_count": 573.5796875, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9774008333683014, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.827823548926972e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.16118774414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0516693115234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3420, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30031040608882903, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.13407388478517532, + "grad_norm": 0.00038561245310120285, + "kept_token_count": 572.4328125, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.976138311624527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.320985991507768e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 183.76156463623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.87274169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3430, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.30158979892730714, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.14147528260946274, + "grad_norm": 0.00033720958163030446, + "kept_token_count": 571.240625, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9758274734020234, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.442414884688332e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.99239959716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.46409301757814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3440, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.299074849486351, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.1351974166929722, + "grad_norm": 0.000353198847733438, + "kept_token_count": 573.7796875, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9763576745986938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.23530911677517e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.91033782958985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.94727478027346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3450, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.29969445765018465, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.1392744928598404, + "grad_norm": 0.00039138871943578124, + "kept_token_count": 572.759375, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9779938697814942, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.596158149885014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.47287292480468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.58140869140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3460, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.30161143839359283, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.1303740605711937, + "grad_norm": 0.0004488560080062598, + "kept_token_count": 571.590625, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9764462828636169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.200686981785111e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.16233673095704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.32179260253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3470, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.29937152862548827, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.14311837106943132, + "grad_norm": 0.0005184969049878418, + "kept_token_count": 573.5734375, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9768201768398285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.054644469870254e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.10323181152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3512817382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3480, + "valid_token_count": 818.6578125 + }, + { + "actual_token_mask_ratio": 0.29983041286468504, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.12413489446043968, + "grad_norm": 0.0004058822523802519, + "kept_token_count": 572.94375, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9770367741584778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.970024282461964e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.12267150878907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3490, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30032464265823366, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.12339229807257653, + "grad_norm": 0.0003852884692605585, + "kept_token_count": 572.5890625, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9771578788757325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.922723936848343e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.26669616699218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10162658691405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3500, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2999959349632263, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.1356051631271839, + "grad_norm": 0.00039514043601229787, + "kept_token_count": 572.7390625, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9765781700611115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.149172401521355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.0744842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5137237548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3510, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30004618167877195, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.12372775450348854, + "grad_norm": 0.00037003474426455796, + "kept_token_count": 572.8015625, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9772415280342102, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.890051176422276e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.22532501220704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9491943359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3520, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29958500266075133, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.1374598741531372, + "grad_norm": 0.0004014665319118649, + "kept_token_count": 573.2140625, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9766086339950562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.13727177248802e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.07446746826173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.89329528808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29965382218360903, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.12890284284949302, + "grad_norm": 0.0003973661223426461, + "kept_token_count": 573.3953125, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9770901143550873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.949188704718836e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.720361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1189819335938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3540, + "valid_token_count": 818.73125 + }, + { + "actual_token_mask_ratio": 0.30046868324279785, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.13467366322875024, + "grad_norm": 0.00045673426939174533, + "kept_token_count": 572.3890625, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9765964329242707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.14203563297633e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.73184509277343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0530700683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3550, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3009104609489441, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.13350128903985023, + "grad_norm": 0.0003484539920464158, + "kept_token_count": 571.9640625, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9768978178501129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.024310566019267e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.15795135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4508026123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3560, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30020955502986907, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.12565554603934287, + "grad_norm": 0.00037239925586618483, + "kept_token_count": 572.5671875, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9776407659053803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.734099974390119e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.16700592041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1177673339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30095742344856263, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.13863620460033416, + "grad_norm": 0.0004021997156087309, + "kept_token_count": 572.1859375, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.976462185382843, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.19447535125073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.63091125488282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.71739501953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3580, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.3003818243741989, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.1432029776275158, + "grad_norm": 0.0003520310274325311, + "kept_token_count": 572.55625, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9770544528961181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.96311663382221e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.67117309570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0022033691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3590, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.3005345791578293, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.13130167052149772, + "grad_norm": 0.00035054347245022655, + "kept_token_count": 572.3796875, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9776037216186524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.748567488510161e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.3101058959961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68226623535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3600, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29951211810112, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.1287343516945839, + "grad_norm": 0.00039450329495593905, + "kept_token_count": 573.0125, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9776087701320648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.746593739488162e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.42135772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0214416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3610, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3006180226802826, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.13171194344758988, + "grad_norm": 0.00041607426828704774, + "kept_token_count": 571.9796875, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.977787172794342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.676906145410612e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.35186462402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3534912109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3620, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.29975145161151884, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.1343156322836876, + "grad_norm": 0.00030975864501670003, + "kept_token_count": 572.8578125, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9777421474456787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.694497519172728e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.67495880126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4073944091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3630, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30035113990306855, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.12388032525777817, + "grad_norm": 0.000399319309508428, + "kept_token_count": 572.7109375, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.977603280544281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.748742766329087e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.2292221069336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8603790283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3640, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.2996224910020828, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.1392678789794445, + "grad_norm": 0.00036092346999794245, + "kept_token_count": 573.0375, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9769213020801544, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.015139512484893e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.42015991210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3771514892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3650, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.3006940811872482, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.13001641929149627, + "grad_norm": 0.0004086991539224982, + "kept_token_count": 572.225, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9772885739803314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.87166999746114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.1170623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4113342285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.300078347325325, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.1307630270719528, + "grad_norm": 0.0004608201561495662, + "kept_token_count": 572.7453125, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9765461921691895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.1616631107172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.97808990478515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7803009033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3670, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3002575904130936, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.12568695470690727, + "grad_norm": 0.00045292574213817716, + "kept_token_count": 572.834375, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9761021256446838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 9.335126087535172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.85135040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5483428955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3680, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.2993044823408127, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.13718232735991479, + "grad_norm": 0.00043062528129667044, + "kept_token_count": 573.3484375, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9770965576171875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.946675952756778e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.64554595947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4209411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3690, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30011440217494967, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.12571242749691008, + "grad_norm": 0.000420195545302704, + "kept_token_count": 572.9359375, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9785031735897064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.397215424338356e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.16370086669923, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.27101440429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3700, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.30031241476535797, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.13472337797284126, + "grad_norm": 0.00036096348776482046, + "kept_token_count": 572.4703125, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9770973205566407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.946378729888239e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.62471466064454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1704956054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3710, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3003495275974274, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.13073959723114967, + "grad_norm": 0.00037515131407417357, + "kept_token_count": 572.596875, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9771208345890046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.937194434111007e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.3847885131836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.67102966308596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3720, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.29973405301570893, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.12310578972101212, + "grad_norm": 0.0004701344878412783, + "kept_token_count": 573.0, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9772754669189453, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.876789215719327e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.09886932373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8981994628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3730, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30025230050086976, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.13085472211241722, + "grad_norm": 0.0004364855703897774, + "kept_token_count": 572.6234375, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9773957788944244, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.829790749587119e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.8483688354492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70204772949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3740, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.3003810286521912, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.14207931309938432, + "grad_norm": 0.0003078164882026613, + "kept_token_count": 572.178125, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9785677671432496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.371990334126167e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.69604034423827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.51003112792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3750, + "valid_token_count": 817.840625 + }, + { + "actual_token_mask_ratio": 0.30045440793037415, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.13621295019984245, + "grad_norm": 0.00030908710323274136, + "kept_token_count": 572.4109375, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9774295032024384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.816612826194614e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 206.00874786376954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7110870361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3760, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30011268258094786, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.13555112928152085, + "grad_norm": 0.0003550265682861209, + "kept_token_count": 572.5265625, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790316462516785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.190788284991867e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.23990783691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4919097900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3770, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.29958145022392274, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.13347167819738387, + "grad_norm": 0.00031814511748962104, + "kept_token_count": 573.23125, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9780863881111145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.56002516229637e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.92359771728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6535369873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3780, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29944940507411955, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.12646338939666749, + "grad_norm": 0.000407516083214432, + "kept_token_count": 573.1390625, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9797523021697998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.909275082056411e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.86134033203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.23383178710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3790, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.3000263273715973, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.1377827726304531, + "grad_norm": 0.0003562782076187432, + "kept_token_count": 572.7015625, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9784947752952575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.400497608818114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.51139373779296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.42076721191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3800, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2996886819601059, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.13022648245096208, + "grad_norm": 0.00038447973201982677, + "kept_token_count": 572.8671875, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9774419486522674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.811753868940287e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.62472381591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.31493225097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3810, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2995910465717316, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.13355168104171752, + "grad_norm": 0.0003727836301550269, + "kept_token_count": 573.1234375, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788391172885895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.265987780760042e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.996484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.52434692382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3820, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29971290230751035, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.1286648131906986, + "grad_norm": 0.0005099981208331883, + "kept_token_count": 573.1484375, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9785410463809967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.382422820432111e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.28770141601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0555938720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3830, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.29892171919345856, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.12978905737400054, + "grad_norm": 0.00028591969748958945, + "kept_token_count": 573.6390625, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9781516075134278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.534539956599474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.51259765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9646331787109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3840, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2987226575613022, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.1303161323070526, + "grad_norm": 0.0004897013423033059, + "kept_token_count": 573.8921875, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9781437277793884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.537621542927809e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.76344299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4092681884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3850, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2991383969783783, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.13356260433793068, + "grad_norm": 0.000361567857908085, + "kept_token_count": 573.3015625, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9771728038787841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.916888255043887e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.74563293457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61842346191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3860, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30028067231178285, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.13318901285529136, + "grad_norm": 0.0003527382796164602, + "kept_token_count": 572.5546875, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790129780769348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.198071009246632e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.57003936767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0956329345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3870, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.29919897615909574, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.12463764250278472, + "grad_norm": 0.000309545430354774, + "kept_token_count": 573.3921875, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793299198150635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.074272045632824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.1176330566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0266479492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3880, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.3001848220825195, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.13660700395703315, + "grad_norm": 0.0004153034824412316, + "kept_token_count": 572.6390625, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9779944598674775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.595932886237278e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.0119430541992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7335479736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3890, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.29946279227733613, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.13186948001384735, + "grad_norm": 0.0003666483098641038, + "kept_token_count": 573.35625, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9775839984416962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.75627294590231e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.797265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.692724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3900, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.30085954666137693, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.14103865772485732, + "grad_norm": 0.0004395712458062917, + "kept_token_count": 572.1703125, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9779681742191315, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.606201881775633e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.77646026611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.47076416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3910, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.3011267751455307, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.1420359916985035, + "grad_norm": 0.00038008231786079705, + "kept_token_count": 571.8609375, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9783810198307037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.444932536804118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.4137710571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7653350830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3920, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2992582321166992, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.1307193748652935, + "grad_norm": 0.0004535709449555725, + "kept_token_count": 573.034375, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9796281456947327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.95777072198689e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.92621765136718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2664001464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3930, + "valid_token_count": 817.759375 + }, + { + "actual_token_mask_ratio": 0.3003665566444397, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.13656492829322814, + "grad_norm": 0.00046607578406110406, + "kept_token_count": 572.409375, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.978268563747406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.488866951665841e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.4437744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.31865844726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3002583414316177, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.1344378784298897, + "grad_norm": 0.00040493629057891667, + "kept_token_count": 572.5359375, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791171252727509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.157391857821494e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.53666229248046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7488677978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3950, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.3001114398241043, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.13354749530553817, + "grad_norm": 0.00045735909952782094, + "kept_token_count": 572.6140625, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9779327571392059, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.620039734523744e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.8318313598633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.92418212890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3960, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.30013791620731356, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.1353535421192646, + "grad_norm": 0.00033700073254294693, + "kept_token_count": 572.6796875, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.978530764579773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.386432236875407e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.0345031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1039154052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3005355179309845, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.14226943999528885, + "grad_norm": 0.00031645296257920563, + "kept_token_count": 572.4453125, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788691341876984, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.254267304437235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.78168792724608, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.66404113769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3980, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.2999899089336395, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.14151096120476722, + "grad_norm": 0.0004643313295673579, + "kept_token_count": 572.728125, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9784309267997742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.425434862147085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.9727813720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2144256591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3990, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2997912883758545, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.13323840275406837, + "grad_norm": 0.00043218882638029754, + "kept_token_count": 572.6890625, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790239751338958, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.193783141905442e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.50955963134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8734161376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4000, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.2997392386198044, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.13746600821614266, + "grad_norm": 0.0003762274282053113, + "kept_token_count": 572.903125, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793851852416993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.05268093245104e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.34825592041017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2003662109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4010, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3004727721214294, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.15011695995926858, + "grad_norm": 0.0003862055018544197, + "kept_token_count": 572.60625, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788873553276062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.247143850894644e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.87989654541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7581756591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4020, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.30027436912059785, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.1341738983988762, + "grad_norm": 0.00038493884494528174, + "kept_token_count": 572.7109375, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9802021265029908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.733569218544289e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.08079376220704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82972106933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4030, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.30046922862529757, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.12790612429380416, + "grad_norm": 0.00037641069502569735, + "kept_token_count": 572.225, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791150987148285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.158184209605679e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.5509521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7524719238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4040, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.30024160742759703, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.12658654749393464, + "grad_norm": 0.0003863586753141135, + "kept_token_count": 572.69375, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790938973426819, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.166469633579254e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.36272735595702, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01715087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.2989754229784012, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.12798314690589904, + "grad_norm": 0.0004160354728810489, + "kept_token_count": 573.5015625, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9803804993629456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.663886426598765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 212.86138916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.72081298828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4060, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2994843155145645, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.13622259348630905, + "grad_norm": 0.0004094239557161927, + "kept_token_count": 573.2015625, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9779688775539398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.605925322626717e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.8938186645508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4070, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30028451085090635, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.13610579669475556, + "grad_norm": 0.0004609439056366682, + "kept_token_count": 572.7, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791004240512848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.163918828358873e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.12827758789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4171142578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4080, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.3010725110769272, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.13334926441311837, + "grad_norm": 0.0004861532070208341, + "kept_token_count": 571.628125, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9782643258571625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.490519685437903e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.49535064697267, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70400085449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4090, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.30117841362953185, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.13521892800927163, + "grad_norm": 0.0004280037828721106, + "kept_token_count": 571.76875, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9792538762092591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.10397817986086e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.0892532348633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.30983276367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4100, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.3002904802560806, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.13306519910693168, + "grad_norm": 0.00033928165794350207, + "kept_token_count": 572.85, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9774800896644592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.796862384770065e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.69046936035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8565155029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4110, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.3005658596754074, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.13539966121315955, + "grad_norm": 0.00037262155092321336, + "kept_token_count": 572.2609375, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9795923829078674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.971751329023391e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.33150939941407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.35300598144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4120, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.299792742729187, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.12175936251878738, + "grad_norm": 0.0003977067244704813, + "kept_token_count": 572.9109375, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9798131465911866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.885508457547986e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.1488235473633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9880676269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4130, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.29919542372226715, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.14680152386426926, + "grad_norm": 0.0003285222628619522, + "kept_token_count": 573.51875, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788178026676178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.274314022855833e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.87858123779296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0151733398437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4140, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2991976082324982, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.13872235119342805, + "grad_norm": 0.0003790365008171648, + "kept_token_count": 573.3265625, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.978453928232193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.41645167383831e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.19497375488282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.24695739746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4150, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.29974470734596254, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.1473669946193695, + "grad_norm": 0.0003883266472257674, + "kept_token_count": 572.9578125, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788999795913697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.242220428655855e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.98035278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1603759765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4160, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30140950083732604, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.14695237576961517, + "grad_norm": 0.0003433894016779959, + "kept_token_count": 571.459375, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.978793466091156, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.283822462544777e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.16824493408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0029998779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4170, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.30115412175655365, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.13818552792072297, + "grad_norm": 0.00043382254079915583, + "kept_token_count": 571.809375, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793688237667084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.05906776804477e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.02169952392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6585662841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4180, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30073525607585905, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.13308771401643754, + "grad_norm": 0.0003189223643857986, + "kept_token_count": 572.3953125, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9789839327335358, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.209416992031038e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.97929840087892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3410888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4190, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.30028151273727416, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.1252038970589638, + "grad_norm": 0.00041272275848314166, + "kept_token_count": 572.48125, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791963875293732, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.126433604047634e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.0869567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4915496826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4200, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30040682256221773, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.12812932953238487, + "grad_norm": 0.0004022795183118433, + "kept_token_count": 572.396875, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790846526622772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.170076762326062e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.9530807495117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.29644470214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4210, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29980679750442507, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.12625850811600686, + "grad_norm": 0.00045968961785547435, + "kept_token_count": 573.09375, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9785459816455842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.380497092730366e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.44935913085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2143218994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4220, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.29962781369686126, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.1308380275964737, + "grad_norm": 0.0006164056248962879, + "kept_token_count": 573.3578125, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9788377523422241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.266520671895705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.8384262084961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8320739746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4230, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.29998440444469454, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.1387266531586647, + "grad_norm": 0.00029879569774493575, + "kept_token_count": 572.8484375, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810439586639405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.404720381600782e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.0772262573242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2127655029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4240, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29827878475189207, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.12821516022086143, + "grad_norm": 0.0005078324466012418, + "kept_token_count": 574.2109375, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.979975163936615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.822214465704746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.0223159790039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.18695373535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4250, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30013648271560667, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.13281850516796112, + "grad_norm": 0.0003543723141774535, + "kept_token_count": 572.6859375, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9796639263629914, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.94380292063579e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.35723571777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1126251220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4260, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3005231976509094, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.12804253473877908, + "grad_norm": 0.0003603834775276482, + "kept_token_count": 572.303125, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9794833362102509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.014340637600981e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.4779495239258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0437835693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4270, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2996793150901794, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.13656176552176474, + "grad_norm": 0.0004060724168084562, + "kept_token_count": 573.096875, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9795367240905761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.993481995072216e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.11168518066407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06227416992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4280, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3010114192962646, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.1360725112259388, + "grad_norm": 0.00045393555774353445, + "kept_token_count": 572.0375, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9787281215190887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.309345575980842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.81651000976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6238250732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4290, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.29979618191719054, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.14076007902622223, + "grad_norm": 0.0004118549113627523, + "kept_token_count": 572.8625, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9797878324985504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.89539801189676e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.75553131103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.62042236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4300, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2991144746541977, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.14050800055265428, + "grad_norm": 0.0004056085308548063, + "kept_token_count": 573.6875, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9781361401081086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.540586495655588e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.6965072631836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.69716186523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4310, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.29993350207805636, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.12871826589107513, + "grad_norm": 0.00044045023969374597, + "kept_token_count": 572.771875, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791027665138244, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.162999656633474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.56976623535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5773468017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4320, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2996249824762344, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.13803989812731743, + "grad_norm": 0.00043236007331870496, + "kept_token_count": 573.0625, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790332436561584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.19016157038277e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.27820434570313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4919769287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4330, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.29990417659282687, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.14522882252931596, + "grad_norm": 0.00042279850458726287, + "kept_token_count": 572.565625, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9790122747421265, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.198347495635971e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.60929412841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4970458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4340, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.30043599009513855, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.143735521286726, + "grad_norm": 0.00039050637860782444, + "kept_token_count": 572.384375, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9794253945350647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.036974177230149e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.4356887817383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4648193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.299537381529808, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.13831151127815247, + "grad_norm": 0.000492945546284318, + "kept_token_count": 573.0125, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9792127192020417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.120049169519916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.31148986816407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.41859436035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4360, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.29965507984161377, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.1278818890452385, + "grad_norm": 0.00039542754529975355, + "kept_token_count": 573.1125, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9806428134441376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.561419333796948e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.90622253417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3120422363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4370, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.29973560869693755, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.14387641325592995, + "grad_norm": 0.0003555933653842658, + "kept_token_count": 572.971875, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.978915399312973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.236193243646995e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.62768096923827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2578552246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4380, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29905771017074584, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.1329694904386997, + "grad_norm": 0.000414699490647763, + "kept_token_count": 573.725, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9801876604557037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.739220600342379e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.68531341552733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4916625976563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4390, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.2999255955219269, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.14438605308532715, + "grad_norm": 0.0003763653221540153, + "kept_token_count": 573.0625, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9789138555526733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.236797584686429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.35872802734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3789794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4400, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.30049207210540774, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.14521378129720688, + "grad_norm": 0.00037341410643421113, + "kept_token_count": 572.603125, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9792438745498657, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.107879257295281e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.672314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.40016479492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.2999092102050781, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.14343623593449592, + "grad_norm": 0.0003113674174528569, + "kept_token_count": 573.13125, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9797830939292907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.897245377535001e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.13988800048827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2360870361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4420, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.30069052875041963, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.14033826515078546, + "grad_norm": 0.0004415308067109436, + "kept_token_count": 572.2171875, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.979806387424469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.888148466008715e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.014990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9392852783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4430, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3004725188016891, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.1303198978304863, + "grad_norm": 0.00033419043757021427, + "kept_token_count": 572.6015625, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9805151462554932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.611291584908031e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.33464508056642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0604583740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4440, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.3001108020544052, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.12736311703920364, + "grad_norm": 0.00029574966174550354, + "kept_token_count": 572.665625, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9798010468482972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.890236229286529e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.34320526123048, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6093444824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29972149431705475, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.12889542058110237, + "grad_norm": 0.0004138484364375472, + "kept_token_count": 572.7, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9804453492164612, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.63855190598406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.76447143554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7105407714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4460, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.2991479992866516, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.13128964677453042, + "grad_norm": 0.00036961378646083176, + "kept_token_count": 573.44375, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9807792186737061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.508141861762851e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.12848052978515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3457458496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4470, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.3007125288248062, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.136357568949461, + "grad_norm": 0.00035194779047742486, + "kept_token_count": 572.2, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9797470033168793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.911343855084851e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.7217224121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8673156738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4480, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30006223917007446, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.1246808998286724, + "grad_norm": 0.00046776991803199053, + "kept_token_count": 572.5796875, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9795617401599884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.9837124212645e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.50556335449218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3940795898437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4490, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.29972889721393586, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.14234553277492523, + "grad_norm": 0.00047525286208838224, + "kept_token_count": 572.9, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9801102042198181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.76947068516165e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.5722396850586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.88804626464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4500, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30003186166286466, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.1192746877670288, + "grad_norm": 0.00042642850894480944, + "kept_token_count": 572.83125, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793051481246948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.083943066594657e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.4621826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9167541503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4510, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.30041203200817107, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.1226195178925991, + "grad_norm": 0.0004326604539528489, + "kept_token_count": 572.3828125, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793154060840606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.079940089373849e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.95596466064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3370391845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4520, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2993598461151123, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.12460339218378066, + "grad_norm": 0.000437427603174001, + "kept_token_count": 573.1390625, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9802092850208283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.730769211775624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.25625915527343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.83685607910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.29903310239315034, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.12918104752898216, + "grad_norm": 0.00039487567846663296, + "kept_token_count": 573.5359375, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9785467624664307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.380190047319047e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.653759765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.23658142089846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4540, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.3010657489299774, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.1315353661775589, + "grad_norm": 0.0003349041799083352, + "kept_token_count": 571.9328125, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9799857616424561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.818081357982009e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.45979919433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7257019042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4550, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2998850464820862, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.1305128589272499, + "grad_norm": 0.0003770359617192298, + "kept_token_count": 572.7890625, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816227674484252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.178628038673196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.06488952636718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.57142944335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4560, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2989652007818222, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.12585159465670587, + "grad_norm": 0.00046910252422094345, + "kept_token_count": 573.609375, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980129873752594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.761789311189204e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.7260314941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3071044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4570, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2992126613855362, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.13628859668970109, + "grad_norm": 0.0003804508887697011, + "kept_token_count": 573.203125, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9800142586231232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.806950306985528e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.99722900390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.83140258789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4580, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.30060929954051974, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.13379820436239243, + "grad_norm": 0.0004123813705518842, + "kept_token_count": 572.3765625, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9794717013835907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.018884545890614e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.60208282470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.64500732421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4590, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.30015006959438323, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.12621484473347663, + "grad_norm": 0.0003778991231229156, + "kept_token_count": 572.471875, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9808259844779968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.489870477002114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.97270812988282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5802886962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4600, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.3003600984811783, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.13725456222891808, + "grad_norm": 0.0003673651663120836, + "kept_token_count": 572.471875, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9804106414318084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.652112690266221e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.5392639160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.58062133789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4610, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3005871117115021, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.13237657845020295, + "grad_norm": 0.0004012823337689042, + "kept_token_count": 572.16875, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9806067764759063, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.575498384539969e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.61366119384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2869934082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4620, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.30007328987121584, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.13297925591468812, + "grad_norm": 0.0003360111149959266, + "kept_token_count": 572.8109375, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9805481016635895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.598418014822527e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.7126892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0943237304688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4630, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2988436549901962, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.1306046187877655, + "grad_norm": 0.0003809012414421886, + "kept_token_count": 573.80625, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813543379306793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.283478626050056e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.21236267089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77161865234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4640, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30010375678539275, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.12601942494511603, + "grad_norm": 0.00035885171382687986, + "kept_token_count": 572.6203125, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9797874569892884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.895545568317174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.13565063476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6591430664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30017009377479553, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.1344509355723858, + "grad_norm": 0.000469675927888602, + "kept_token_count": 572.8140625, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9808319449424744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.487541151931509e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.44080657958983, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9294067382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4660, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2989058643579483, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.14410757198929786, + "grad_norm": 0.00038276720442809165, + "kept_token_count": 573.6296875, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813801050186157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.273410519701429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.81265716552736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0013824462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4670, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3004498422145844, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.14128593504428863, + "grad_norm": 0.0004842751659452915, + "kept_token_count": 572.5453125, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980465030670166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.630864347447641e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.92074584960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7069580078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4680, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30073759257793425, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.1384745366871357, + "grad_norm": 0.0004386711516417563, + "kept_token_count": 572.2171875, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9807001233100892, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.539031284977681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.2380615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0636291503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4690, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30074972212314605, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.13816029950976372, + "grad_norm": 0.0004972657188773155, + "kept_token_count": 572.059375, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9793153762817383, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.079951730906032e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.03418579101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0242034912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4700, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.3006596177816391, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.12329818233847618, + "grad_norm": 0.00047333334805443883, + "kept_token_count": 572.2875, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980737853050232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.524295724579134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.81533966064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.88849182128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4710, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3003766894340515, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.12660174667835236, + "grad_norm": 0.00038558177766390145, + "kept_token_count": 572.8171875, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9798913657665252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.854956566006876e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.3674575805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.63690795898435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4720, + "valid_token_count": 818.75 + }, + { + "actual_token_mask_ratio": 0.3004558175802231, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.12856353521347047, + "grad_norm": 0.0003067039360757917, + "kept_token_count": 572.246875, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9809509813785553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.441038505930919e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.40086364746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.86773986816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4730, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.3004732966423035, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.12909073010087013, + "grad_norm": 0.00030115703702904284, + "kept_token_count": 572.378125, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817073225975037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.145599302020856e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.52987365722657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0533966064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4740, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.30099970996379855, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.1444212444126606, + "grad_norm": 0.00036539044231176376, + "kept_token_count": 572.121875, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980260694026947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.710690988460556e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.0069778442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7717803955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4750, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.30153925716876984, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.1340853229165077, + "grad_norm": 0.0003468952781986445, + "kept_token_count": 571.4078125, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9806895554065704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.543159154010937e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.28160095214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.73704833984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.3010639399290085, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.13393581956624984, + "grad_norm": 0.0004960994119755924, + "kept_token_count": 571.7890625, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980406528711319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.653714201296679e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.50702819824218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7175598144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4770, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2998631089925766, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.14552430287003518, + "grad_norm": 0.0004212426720187068, + "kept_token_count": 572.9578125, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9799278199672699, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.840716643841006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.23776092529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3282073974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4780, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.30039624571800233, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.14488606229424478, + "grad_norm": 0.0003393793012946844, + "kept_token_count": 572.5375, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9808128535747528, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.495001336792484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.06908416748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.91912536621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4790, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.29927348494529726, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.12295618653297424, + "grad_norm": 0.00037475983845070004, + "kept_token_count": 573.63125, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981809812784195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.105557742761447e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.59328002929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3619354248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4800, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.2997779667377472, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.1371731162071228, + "grad_norm": 0.00042325095273554325, + "kept_token_count": 573.0640625, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817893087863923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.113570900401101e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.55757598876954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6120208740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4810, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.30009484887123106, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.13307221829891205, + "grad_norm": 0.0003562601050361991, + "kept_token_count": 572.7546875, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9812474548816681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.325228725676424e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.14434356689452, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5339813232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4820, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29973213374614716, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.13544553592801095, + "grad_norm": 0.00036212956183589995, + "kept_token_count": 573.146875, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810779511928558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.391442486550658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.90628662109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.51340637207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4830, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.30053127408027647, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.13500563725829123, + "grad_norm": 0.000309611699776724, + "kept_token_count": 572.3734375, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981222003698349, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.33517503249459e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.11421813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36313781738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4840, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.299987730383873, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.12847048342227935, + "grad_norm": 0.0002659499004948884, + "kept_token_count": 572.8859375, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816566526889801, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.165390816226136e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.8056167602539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3492401123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4850, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29972051084041595, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.14460848048329353, + "grad_norm": 0.00031274641514755785, + "kept_token_count": 573.10625, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9815483748912811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.207682865555398e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.19435272216796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.09150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4860, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.3002253770828247, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.14554993212223052, + "grad_norm": 0.00044163939310237765, + "kept_token_count": 572.621875, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9800028562545776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.81139882747084e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.54261169433593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8173400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4870, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30133984088897703, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.14339060112833976, + "grad_norm": 0.00040471923421137035, + "kept_token_count": 571.5828125, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9805949807167054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.580105193483178e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.32428131103515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7521026611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4880, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.29955623745918275, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.12699809446930885, + "grad_norm": 0.00033190581598319113, + "kept_token_count": 573.071875, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981080275774002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.390533792204224e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.11450805664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.49423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4890, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30011657178401946, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.13320533260703088, + "grad_norm": 0.00035223542363382876, + "kept_token_count": 572.78125, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810533404350281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.401057373499498e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.40611877441407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61318359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4900, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3001629710197449, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.14004360660910606, + "grad_norm": 0.0005528704496100545, + "kept_token_count": 572.484375, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9807870149612427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.505082758143544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.52246551513673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3614105224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4910, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.29916236996650697, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.14254266694188117, + "grad_norm": 0.00030580104794353247, + "kept_token_count": 573.2734375, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9791171073913574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.157394040608779e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.8258087158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1768798828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4920, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.2997824490070343, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.13730627968907355, + "grad_norm": 0.0004035016754642129, + "kept_token_count": 573.0140625, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9826683342456818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.770201325707603e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.45781860351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.97840576171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4930, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.2992573767900467, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.12957227006554603, + "grad_norm": 0.00040339634870179, + "kept_token_count": 573.51875, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980494099855423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.619510579388588e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.1578353881836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17624816894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4940, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.29915992021560667, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.1350291259586811, + "grad_norm": 0.0005273426650092006, + "kept_token_count": 573.45625, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813395380973816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.28926133888308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.4025421142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4027069091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29963170886039736, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.1326457977294922, + "grad_norm": 0.0003205333196092397, + "kept_token_count": 573.246875, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9812093555927277, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.340109696087893e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.0995086669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7648162841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4960, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.2989685952663422, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.14008130505681038, + "grad_norm": 0.00034245444112457335, + "kept_token_count": 573.61875, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810252249240875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.412038030452095e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.41273498535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.27940979003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4970, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.29972855746746063, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.14257756024599075, + "grad_norm": 0.00037110026460140944, + "kept_token_count": 572.93125, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810605406761169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.398245797958225e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.13695220947267, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.87843017578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30073761343955996, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.13742267489433288, + "grad_norm": 0.0004151401808485389, + "kept_token_count": 572.11875, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981763631105423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.123602335923351e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.11797332763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.75213928222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4990, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2996354401111603, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.13691724836826324, + "grad_norm": 0.00033165531931445, + "kept_token_count": 573.2890625, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822693347930909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.926067217136734e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.1727523803711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89971618652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5000, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.2999934434890747, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.1379638656973839, + "grad_norm": 0.00041559754754416645, + "kept_token_count": 572.90625, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817987203598022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.109887301339767e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.18952484130858, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7661071777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5010, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2997437655925751, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.1394879177212715, + "grad_norm": 0.0004228274920023978, + "kept_token_count": 572.9921875, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9807019591331482, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.538313511759042e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.91809692382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8884796142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.3000731557607651, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.13179022818803787, + "grad_norm": 0.0002784302632790059, + "kept_token_count": 572.934375, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9814650058746338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.240248451125808e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.88695526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0854064941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5030, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30003519654273986, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.13254085332155227, + "grad_norm": 0.0002810803707689047, + "kept_token_count": 572.890625, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981384027004242, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.271887443494052e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.13766479492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.84898376464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5040, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2996514469385147, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.13197681307792664, + "grad_norm": 0.00037572550354525447, + "kept_token_count": 573.3078125, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9818002700805664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.109286270861049e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.23087463378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.00095520019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5050, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.29953538477420805, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.13179319947957993, + "grad_norm": 0.0003963549097534269, + "kept_token_count": 573.0296875, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9824104309082031, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.870942088426091e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.26461334228514, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.93421630859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5060, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.2994863480329514, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.13710032925009727, + "grad_norm": 0.00035992672201246023, + "kept_token_count": 573.3703125, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813485264778137, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.285748433787376e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.66837158203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5493194580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5070, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2996050626039505, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.13782891258597374, + "grad_norm": 0.00035298115108162165, + "kept_token_count": 573.253125, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820713937282562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.003381797403563e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.7670440673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2995146870613098, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.13442696183919906, + "grad_norm": 0.0003540041798260063, + "kept_token_count": 573.134375, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9819364011287689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.056118374748621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.91235809326173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6784149169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5090, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2989556282758713, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.13699048906564712, + "grad_norm": 0.00027924671303480864, + "kept_token_count": 573.459375, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9815603375434876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.203006607596763e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.1122619628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5340118408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.3003273904323578, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.13726108521223068, + "grad_norm": 0.00030251056887209415, + "kept_token_count": 572.5203125, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981564861536026, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.201244370662608e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.80565643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1755645751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3000413179397583, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.1372300073504448, + "grad_norm": 0.0003789659240283072, + "kept_token_count": 572.684375, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980943763256073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.443864888045937e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.27136993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.47719421386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5120, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.30059029459953307, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.13183025270700455, + "grad_norm": 0.00030975978006608784, + "kept_token_count": 572.4484375, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822430789470673, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.936320496606641e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.85784606933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.43384704589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5130, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.3006731539964676, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.14796196445822715, + "grad_norm": 0.0003740791289601475, + "kept_token_count": 572.275, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9794442594051361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 8.029600721783935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.8978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2277770996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5140, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30034649670124053, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.14507510960102082, + "grad_norm": 0.0003249335568398237, + "kept_token_count": 572.2890625, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9809932231903076, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.42454496503342e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.00640716552735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2134552001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5150, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.2988104909658432, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.1383660577237606, + "grad_norm": 0.00034222641261294484, + "kept_token_count": 573.946875, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9803141415119171, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.68980142311193e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.18189086914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5938659667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5160, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3011604756116867, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.13011279180645943, + "grad_norm": 0.0003134333237539977, + "kept_token_count": 571.8453125, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821063339710235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.989737812546082e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.94775848388673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.42885131835936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5170, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3000173598527908, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.1331436462700367, + "grad_norm": 0.00034623825922608376, + "kept_token_count": 572.7703125, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9812975764274597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.305652325158007e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.35370330810548, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.97898254394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5180, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.29861336648464204, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.14233205169439317, + "grad_norm": 0.00040028756484389305, + "kept_token_count": 573.953125, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9824740648269653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.846088072052226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.5674560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.81781311035155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5190, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.30005075335502623, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.1290663279592991, + "grad_norm": 0.0003624924866016954, + "kept_token_count": 572.79375, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817622125148773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.124154872144573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.11573791503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.42510986328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5200, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30054129660129547, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.129706222563982, + "grad_norm": 0.00031274056527763605, + "kept_token_count": 572.228125, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825465500354766, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.817773937655147e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.60594024658204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.781103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5210, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.29945272505283355, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.12984262630343438, + "grad_norm": 0.0003332927590236068, + "kept_token_count": 573.253125, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823462426662445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.896017875988037e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.12107849121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3419525146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5220, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.299868768453598, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.1354097031056881, + "grad_norm": 0.00032591205672360957, + "kept_token_count": 572.840625, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813092529773713, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.301090226974338e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.2073348999023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.33099060058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5230, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3003602772951126, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.12709958478808403, + "grad_norm": 0.00040815898682922125, + "kept_token_count": 572.4828125, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981330955028534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.292617374332622e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.18720245361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9402374267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5240, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29918305575847626, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.12737202420830726, + "grad_norm": 0.00037175483885221183, + "kept_token_count": 573.728125, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821396172046661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.9767333479831e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.49124908447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.56773376464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5250, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.3001464277505875, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.14925425201654435, + "grad_norm": 0.0002933588402811438, + "kept_token_count": 572.7234375, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.980703592300415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.53767897549551e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.95079040527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.10010681152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5260, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.3002756148576736, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.13054998889565467, + "grad_norm": 0.00030136186978779733, + "kept_token_count": 572.7421875, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821618020534515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.968058878555893e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.2721176147461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2827667236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5270, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.2995563209056854, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.13356466814875603, + "grad_norm": 0.00038920831866562366, + "kept_token_count": 572.9484375, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810949265956879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.384812961390708e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.3521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61268615722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5280, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.2992612898349762, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.13554957062005996, + "grad_norm": 0.00029233749955892563, + "kept_token_count": 573.3390625, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9818402707576752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.093664316926152e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.4073059082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0723175048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5290, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3005262166261673, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.14036708250641822, + "grad_norm": 0.00036201550392434, + "kept_token_count": 572.5453125, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820576250553131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.008759348536842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.24806060791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.26064147949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5300, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2999734699726105, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.13316677361726761, + "grad_norm": 0.0005466778529807925, + "kept_token_count": 572.7078125, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823580741882324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.891397715662606e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.73416900634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.38861389160155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5310, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.299518010020256, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.14075981229543685, + "grad_norm": 0.0004158166120760143, + "kept_token_count": 573.3828125, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98121018409729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.339793155551888e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.6479125976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5324371337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5320, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.2996225953102112, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.12847572788596154, + "grad_norm": 0.0003633593150880188, + "kept_token_count": 573.2359375, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9814602732658386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.242099891300314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.63631286621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6010467529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30066175758838654, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.13242449015378951, + "grad_norm": 0.0003130873665213585, + "kept_token_count": 572.234375, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816723883152008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.159242231864482e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.83353118896486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.66460571289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5340, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30040186941623687, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.12510660365223886, + "grad_norm": 0.0004124657716602087, + "kept_token_count": 572.575, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816844224929809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.154542472562753e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.65538330078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.50160522460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5350, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.2990295022726059, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.12786610648036004, + "grad_norm": 0.000412731955293566, + "kept_token_count": 573.4921875, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813897788524628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.269632696988992e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.26234588623046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.22056884765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5360, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29936625957489016, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.1296358659863472, + "grad_norm": 0.0003086358483415097, + "kept_token_count": 573.2203125, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9810591161251068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.398805173579604e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.93633422851562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.425048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5370, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.30007170140743256, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.14024608656764032, + "grad_norm": 0.0003527358639985323, + "kept_token_count": 572.65, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823219537734985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.905505360919051e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.1289123535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8947509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30037282705307006, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.14036421179771424, + "grad_norm": 0.00039921674760989845, + "kept_token_count": 572.375, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816414713859558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.17132290446898e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.21778411865233, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.30703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.29970650672912597, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.14169977605342865, + "grad_norm": 0.00031829599174670875, + "kept_token_count": 573.29375, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828234791755677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.709604676871094e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.95679779052733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.20814514160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5400, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.30026244521141054, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.1244612142443657, + "grad_norm": 0.00033648309181444347, + "kept_token_count": 572.5671875, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817069828510284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.145731869968585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.99652099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2158264160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30114694535732267, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.13658653870224952, + "grad_norm": 0.00033682325738482177, + "kept_token_count": 571.7484375, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9819714009761811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.042448305583093e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.490283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.73292846679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5420, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.30030748844146726, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.12552590444684028, + "grad_norm": 0.00035334753920324147, + "kept_token_count": 572.5, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821044266223907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.9904757765471e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.41777648925782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71778869628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5430, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2990895390510559, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.12318167388439179, + "grad_norm": 0.00041590406908653677, + "kept_token_count": 573.525, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834788680076599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.453587011492345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.81813659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2561950683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5440, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29966226816177366, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.13473087921738625, + "grad_norm": 0.0003908546350430697, + "kept_token_count": 573.034375, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9815131783485412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.221425548777915e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.7950378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2737335205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5450, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29964558184146883, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.12654586583375932, + "grad_norm": 0.00038294572732411325, + "kept_token_count": 573.1609375, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816868722438812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.153589867812116e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.14442596435546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7864959716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5460, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2997974693775177, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.1368771307170391, + "grad_norm": 0.0003070980601478368, + "kept_token_count": 572.875, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9813185214996338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.297475167433731e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.0271301269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71898498535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5470, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.3001871556043625, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.1355416350066662, + "grad_norm": 0.0004233804065734148, + "kept_token_count": 572.653125, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817196309566498, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.140788802644238e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.891340637207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8548126220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29983055889606475, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.12959541380405426, + "grad_norm": 0.00032323377672582865, + "kept_token_count": 573.0171875, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822777092456818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.922792235855013e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.4215423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.04367980957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5490, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.29981646239757537, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.1420372948050499, + "grad_norm": 0.0002611905802041292, + "kept_token_count": 573.0703125, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9812566578388214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.321639131987467e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.30655975341796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8249084472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.29918157756328584, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.13504930064082146, + "grad_norm": 0.000448309670900926, + "kept_token_count": 573.3546875, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820381045341492, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.016378513071685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.0569732666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0559387207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5510, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.3008736610412598, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.13085651993751526, + "grad_norm": 0.00033309304853901267, + "kept_token_count": 571.996875, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821996092796326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.953300471650437e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.72229461669923, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.69432067871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5520, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.30040432810783385, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.13507230058312417, + "grad_norm": 0.0003667858545668423, + "kept_token_count": 572.4046875, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822796940803528, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.922008578840178e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.7708038330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7589050292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5530, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.29885624051094056, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.14666740149259566, + "grad_norm": 0.00031141063664108515, + "kept_token_count": 573.8953125, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9819186508655549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.063043776724953e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.8254821777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.41127014160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5540, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.29990350306034086, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.1426284544169903, + "grad_norm": 0.0003734918136615306, + "kept_token_count": 572.7265625, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825503826141357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.816275854362174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.90799713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.83868408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5550, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.30049191415309906, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.1502915397286415, + "grad_norm": 0.00042828996083699167, + "kept_token_count": 572.6515625, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9815389752388001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.211355987237766e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.95125579833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.21435546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5560, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.30008081495761874, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.12489326521754265, + "grad_norm": 0.00039491712232120335, + "kept_token_count": 572.9703125, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9827109634876251, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.753546840627677e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.5252166748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.6503143310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5570, + "valid_token_count": 818.6265625 + }, + { + "actual_token_mask_ratio": 0.3002439111471176, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.14474593922495843, + "grad_norm": 0.00024805348948575556, + "kept_token_count": 572.8484375, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821971118450165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.954273594601545e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.81610107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7893035888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5580, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.3004242032766342, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.13646270334720612, + "grad_norm": 0.00039177355938591063, + "kept_token_count": 572.371875, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822140455245971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.947658694116399e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.11363677978517, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4460479736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30110475420951843, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.13446230962872505, + "grad_norm": 0.00029352199635468423, + "kept_token_count": 572.0640625, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825951218605041, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.798795184295158e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.20527801513674, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9623718261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5600, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.3005322128534317, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.14003617763519288, + "grad_norm": 0.0003699843364302069, + "kept_token_count": 572.4921875, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817379832267761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.133617873478215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.7970764160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5634033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2993953973054886, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.1314731977880001, + "grad_norm": 0.00031246247817762196, + "kept_token_count": 573.1484375, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832050263881683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.560558795172255e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.6702087402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8722442626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5620, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.30100844502449037, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.13521242663264274, + "grad_norm": 0.00033002340933308005, + "kept_token_count": 571.9953125, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823472023010253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.895647311466746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.74230499267577, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6282196044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5630, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.299548864364624, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.1394409030675888, + "grad_norm": 0.0002876979415304959, + "kept_token_count": 573.1265625, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9826226830482483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.788032223994377e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.4817749023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9170288085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5640, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29966755509376525, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.14540239050984383, + "grad_norm": 0.0003311327309347689, + "kept_token_count": 573.2546875, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822280466556549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.942182735656389e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.19883880615237, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7224060058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5650, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.29974445700645447, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.12964193597435952, + "grad_norm": 0.0003140739572700113, + "kept_token_count": 572.8609375, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9814789593219757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.23479846783448e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.4463317871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.65308532714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5660, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2997776746749878, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.14202482923865317, + "grad_norm": 0.00036745131365023553, + "kept_token_count": 572.825, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.982798022031784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.719538287143223e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.2088104248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5526977539063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5670, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.30053754448890685, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.1286786027252674, + "grad_norm": 0.0003492528630886227, + "kept_token_count": 572.346875, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820294201374054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.019784643489402e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.8716613769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8672210693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5680, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30146428346633913, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.13737967759370803, + "grad_norm": 0.0003317477530799806, + "kept_token_count": 571.484375, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9818552911281586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.087796402629464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.19469299316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.939990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2999423384666443, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.12300340235233306, + "grad_norm": 0.00037273034104146063, + "kept_token_count": 572.9046875, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823673963546753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.887753625051118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.0722290039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0715393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5700, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2993709951639175, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.1355176277458668, + "grad_norm": 0.00033689121482893825, + "kept_token_count": 573.5203125, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9818031430244446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.10816642822465e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.77655639648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7337921142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5710, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.30042791962623594, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.12901876494288445, + "grad_norm": 0.00030242427601478994, + "kept_token_count": 572.478125, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828726708889007, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.690382651868277e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.89231567382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1135589599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2998658686876297, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.1443801201879978, + "grad_norm": 0.0004049011622555554, + "kept_token_count": 572.753125, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9814509272575378, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.245750675792806e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.57877349853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5447662353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5730, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3000468909740448, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.14052910804748536, + "grad_norm": 0.0003118084277957678, + "kept_token_count": 572.6375, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820181906223298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.02416080457624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.14111175537107, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8546142578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2996956557035446, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.12919702008366585, + "grad_norm": 0.0002807565324474126, + "kept_token_count": 573.1515625, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9826462864875793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.77881354931742e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.5885955810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3160095214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5750, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2995506763458252, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.12734211906790732, + "grad_norm": 0.00036063327570445836, + "kept_token_count": 572.959375, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831135332584381, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.596296734642237e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.7803894042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0781677246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5760, + "valid_token_count": 817.9921875 + }, + { + "actual_token_mask_ratio": 0.2999082148075104, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.1320325955748558, + "grad_norm": 0.00028199382359161973, + "kept_token_count": 572.7546875, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837827146053314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.334895369946026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.2503128051758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.35052185058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5770, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3001686006784439, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.14007959738373757, + "grad_norm": 0.00027100491570308805, + "kept_token_count": 572.5859375, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830200672149658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.632795848418027e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.3017333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77687683105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5780, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2999101191759109, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.1334371969103813, + "grad_norm": 0.00031355477403849363, + "kept_token_count": 573.0515625, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835551023483277, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.423809136322234e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.67802276611326, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4316772460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5790, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.299219411611557, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.13670828714966773, + "grad_norm": 0.00037163871456868947, + "kept_token_count": 573.3203125, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825674772262574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.809593542129732e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.1486053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9603332519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5800, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.3003047287464142, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.13154637813568115, + "grad_norm": 0.0004243883886374533, + "kept_token_count": 572.4359375, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834010124206543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.483998149633407e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.1066436767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.80909118652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5810, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3001984149217606, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.12812643945217134, + "grad_norm": 0.00039113545790314674, + "kept_token_count": 572.5484375, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9826647400856018, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.771604748792014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.21728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.09873352050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5820, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2983197718858719, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.13668466359376907, + "grad_norm": 0.00026829628041014075, + "kept_token_count": 574.0328125, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823841094970703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.88122116116574e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.2072357177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.00413513183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5830, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2995265513658524, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.12281433418393135, + "grad_norm": 0.00042396984645165503, + "kept_token_count": 573.1859375, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836969614028931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.368387730617541e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.2611785888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6303741455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5840, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30105075240135193, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.13487769663333893, + "grad_norm": 0.00039967300835996866, + "kept_token_count": 571.8171875, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822394728660584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.937719699635636e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.83202362060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4920593261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5850, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.30061730444431306, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.1375575505197048, + "grad_norm": 0.00028337843832559884, + "kept_token_count": 572.0625, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837979257106781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.32895105809439e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.7634750366211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.11787109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5860, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.29873658120632174, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.12662107199430467, + "grad_norm": 0.00041867801337502897, + "kept_token_count": 573.7375, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.981682938337326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.155123166739941e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.16069946289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9748931884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5870, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.2991157501935959, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.13705503568053246, + "grad_norm": 0.0003162278444506228, + "kept_token_count": 573.403125, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9817227482795715, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.139570006984287e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.9447647094727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9147186279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5880, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.29995664954185486, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.12971999272704124, + "grad_norm": 0.00034434942062944174, + "kept_token_count": 572.990625, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98286452293396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.693567265756429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.9615814208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4724975585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5890, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2996371418237686, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.13162530362606048, + "grad_norm": 0.0002840919769369066, + "kept_token_count": 573.1078125, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830565929412842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.618534498556983e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.6201965332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.39824829101565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5900, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.30118287801742555, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.13360918834805488, + "grad_norm": 0.00033697468461468816, + "kept_token_count": 571.6234375, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836292445659638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.394853226083796e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.13518981933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.13115844726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.29936901926994325, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.13750529289245605, + "grad_norm": 0.00028267965535633266, + "kept_token_count": 573.3703125, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821933627128601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.955737699172459e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.2492645263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0480194091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5920, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.29971216022968294, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.13216005340218545, + "grad_norm": 0.00037305449950508773, + "kept_token_count": 573.13125, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823101818561554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.910103911650367e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.2808563232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4298797607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5930, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2996489703655243, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.13326310291886329, + "grad_norm": 0.00034829304786399007, + "kept_token_count": 573.0671875, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98243368268013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.861858200863935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.32770080566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.14910278320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5940, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.3003306806087494, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.14082370474934577, + "grad_norm": 0.0004361650499049574, + "kept_token_count": 572.5953125, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9826279163360596, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.785987461626064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.13246154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.85850524902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5950, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.29962382912635804, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.13155104741454124, + "grad_norm": 0.00025940375053323805, + "kept_token_count": 573.2984375, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828887403011322, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.684103464067448e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.0281005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4693969726562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5960, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.2988902986049652, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.13196857422590255, + "grad_norm": 0.00027029847842641175, + "kept_token_count": 573.684375, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9838406324386597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.312269724730867e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.6981964111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1454193115234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5970, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3000765025615692, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.1185996949672699, + "grad_norm": 0.0003214450553059578, + "kept_token_count": 572.915625, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835933387279511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.408868612197693e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.96978454589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.780029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5980, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3006038010120392, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.1381395563483238, + "grad_norm": 0.0003396525280550122, + "kept_token_count": 572.4265625, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839913845062256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.25338336249115e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.45664978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9264434814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5990, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.3000475764274597, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.13077352046966553, + "grad_norm": 0.0004771128296852112, + "kept_token_count": 572.99375, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825599551200866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.81253441143781e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.47473907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.89869689941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6000, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.2991271883249283, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.131230628490448, + "grad_norm": 0.0003233837487641722, + "kept_token_count": 573.2953125, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983400309085846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.484273435489741e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.4315490722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6233215332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6010, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.30030129551887513, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.1398371808230877, + "grad_norm": 0.00032536103390157223, + "kept_token_count": 572.54375, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9829949140548706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.642632906732615e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.6725738525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3390777587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6020, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3007150858640671, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.14569003507494926, + "grad_norm": 0.0003296108916401863, + "kept_token_count": 572.1078125, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822375535964966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.938467631698587e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.08318481445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1827850341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6030, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.3006776928901672, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.12657477110624313, + "grad_norm": 0.0003536277508828789, + "kept_token_count": 571.953125, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833129167556762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.518408263218589e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.87281799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.69032592773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6040, + "valid_token_count": 817.86875 + }, + { + "actual_token_mask_ratio": 0.30010545551776885, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.131365517526865, + "grad_norm": 0.0002551418438088149, + "kept_token_count": 572.79375, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831709742546082, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.573859282070771e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.75986633300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0061462402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6050, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.30019657909870145, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.13460009098052977, + "grad_norm": 0.0004426431842148304, + "kept_token_count": 572.734375, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9824089407920837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.871528894407675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.36736755371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.85838623046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6060, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29995336532592776, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.13188282027840614, + "grad_norm": 0.0003107430529780686, + "kept_token_count": 573.03125, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822228431701661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.944219094293658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.58662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3518890380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6070, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.2996059626340866, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.13839516863226892, + "grad_norm": 0.00032675903639756143, + "kept_token_count": 572.8875, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9827615559101105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.733786067343317e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.41216735839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5437713623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6080, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.300602450966835, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.13575698658823968, + "grad_norm": 0.0003541086916811764, + "kept_token_count": 572.1671875, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9827969133853912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.71998415782582e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.6261291503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3789459228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6090, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2997815012931824, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.1338159918785095, + "grad_norm": 0.00037909092498011887, + "kept_token_count": 573.109375, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837303698062897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.355338155117352e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.7391357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6341125488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6100, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3002971917390823, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.13726069182157516, + "grad_norm": 0.0002941617858596146, + "kept_token_count": 572.99375, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825195074081421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.828331061115023e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.40509033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6852996826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6110, + "valid_token_count": 818.9234375 + }, + { + "actual_token_mask_ratio": 0.3000397562980652, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.14331163465976715, + "grad_norm": 0.00029960813117213547, + "kept_token_count": 572.8125, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832345187664032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.549036552314647e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.51157836914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.49140014648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6120, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.300248321890831, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.14101577326655387, + "grad_norm": 0.00031556744943372905, + "kept_token_count": 572.4078125, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830400228500367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.625013083976228e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.1226043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3380462646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6130, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.29975247085094453, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.13809677436947823, + "grad_norm": 0.00039379479130730033, + "kept_token_count": 572.8578125, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984073406457901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.221345465746708e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.3178344726563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2999603271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6140, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.3003768682479858, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.12357928901910782, + "grad_norm": 0.00031848580692894757, + "kept_token_count": 572.5046875, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837942600250245, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.330392934614793e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.16490936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1655334472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6150, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.3001956850290298, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.13183821737766266, + "grad_norm": 0.000287205126369372, + "kept_token_count": 572.503125, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9820752203464508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.00188447808614e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.6498657226563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2522247314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3003461420536041, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.12815046310424805, + "grad_norm": 0.0002840604283846915, + "kept_token_count": 572.490625, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837524056434631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.346735863189678e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.0128143310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.5262420654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6170, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29958725571632383, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.127823606133461, + "grad_norm": 0.00038659770507365465, + "kept_token_count": 573.23125, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821523666381836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.971741750021465e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.49529724121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4997222900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6180, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.2997789651155472, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.13110610842704773, + "grad_norm": 0.00027237599715590477, + "kept_token_count": 572.8828125, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837904810905457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.331862859951798e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.7272583007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5249481201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6190, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2997601479291916, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.13616390898823738, + "grad_norm": 0.00034325229353271425, + "kept_token_count": 573.046875, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833042323589325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.521800314658321e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.09955139160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1071716308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6200, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.2997569113969803, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.13794463947415353, + "grad_norm": 0.0003002055163960904, + "kept_token_count": 572.890625, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833924293518066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.487346909125335e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.1831817626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.14674682617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6210, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.3002499908208847, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.12771469503641128, + "grad_norm": 0.0002689013199415058, + "kept_token_count": 572.565625, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831780135631562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.571104895556346e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.23167114257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.27223815917966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6220, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.299692639708519, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.13329598754644395, + "grad_norm": 0.00043159310007467866, + "kept_token_count": 573.1375, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9822534441947937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.932264514034615e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.13226318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42226257324216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2996347337961197, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.1456870399415493, + "grad_norm": 0.0003141560300718993, + "kept_token_count": 572.88125, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843630611896514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.108192101237365e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.5214385986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.10767517089846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6240, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.30032926201820376, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.13181351646780967, + "grad_norm": 0.0003263496037106961, + "kept_token_count": 572.615625, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843628942966461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.108264606154989e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.92693481445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8261779785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6250, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.29984118938446047, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.13155683726072312, + "grad_norm": 0.0003953265550080687, + "kept_token_count": 572.9890625, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828434586524963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.701795537082944e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.6836730957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1558135986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6260, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2996376484632492, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.12221218049526214, + "grad_norm": 0.00033590878592804074, + "kept_token_count": 572.9921875, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840770781040191, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.219904571480583e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.3863098144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4085388183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6270, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30009508728981016, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.12059684991836547, + "grad_norm": 0.0003591194690670818, + "kept_token_count": 572.540625, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839763045310974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.259270885493607e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.3460998535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3081970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6280, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2998013705015182, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.13582468256354333, + "grad_norm": 0.00031849718652665615, + "kept_token_count": 573.18125, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832000195980072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.562513372045941e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.4483306884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9582000732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6290, + "valid_token_count": 818.596875 + }, + { + "actual_token_mask_ratio": 0.29993281960487367, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.13121864721179008, + "grad_norm": 0.0004964454565197229, + "kept_token_count": 572.7921875, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9838628053665162, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.303618029050995e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.8991638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2916259765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6300, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2996491551399231, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.12087503522634506, + "grad_norm": 0.00031666879658587277, + "kept_token_count": 573.146875, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983981454372406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.257265995373018e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.4633514404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1481140136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6310, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2994519233703613, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.13259799405932426, + "grad_norm": 0.00030597130535170436, + "kept_token_count": 573.246875, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831604301929474, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.577975073014386e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.21685180664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8809448242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6320, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2994570881128311, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.13868600875139236, + "grad_norm": 0.00037427619099617004, + "kept_token_count": 573.0796875, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9825770676136016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.805849334341474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.3529510498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.12347717285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6330, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.3007943481206894, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.13710299804806708, + "grad_norm": 0.0002777735935524106, + "kept_token_count": 572.328125, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833771705627441, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.493313703685999e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.6310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.39854431152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6340, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.2999917894601822, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.13846408277750016, + "grad_norm": 0.00039356868364848197, + "kept_token_count": 572.7546875, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840538084506989, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.228996353456751e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.80585021972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2636749267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2994701027870178, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.13764237314462663, + "grad_norm": 0.0003192908188793808, + "kept_token_count": 573.2234375, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9823494195938111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.894773832755163e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.3696319580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4596466064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6360, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30145072042942045, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.13284454122185707, + "grad_norm": 0.000371801812434569, + "kept_token_count": 571.55625, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98378484249115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.334068057185504e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.54144287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61192626953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6370, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30003719627857206, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.13757565394043922, + "grad_norm": 0.00033941399306058884, + "kept_token_count": 573.09375, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9824138343334198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.869613680464682e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.02998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.90850830078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6380, + "valid_token_count": 818.7484375 + }, + { + "actual_token_mask_ratio": 0.30060866475105286, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.13445921540260314, + "grad_norm": 0.00028946108068339527, + "kept_token_count": 572.2671875, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833662450313568, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.497582617157605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.9680206298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.05259094238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6390, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3003244936466217, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.1368352308869362, + "grad_norm": 0.00033822821569629014, + "kept_token_count": 572.4609375, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9827533006668091, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.737010844517499e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.78355102539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0464630126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6400, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2995702177286148, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.14287057220935823, + "grad_norm": 0.00041040318319574, + "kept_token_count": 573.1375, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837326526641845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.354454308166169e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.09251098632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.38421630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.3004025310277939, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.127397770434618, + "grad_norm": 0.00029579305555671453, + "kept_token_count": 572.334375, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835016191005707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.444694263336714e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.9868438720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7142822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6420, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.29974467754364015, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.13786037862300873, + "grad_norm": 0.0002802644739858806, + "kept_token_count": 572.9390625, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983199542760849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.562696034961846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.9526885986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3658508300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6430, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.3006410479545593, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.12931784614920616, + "grad_norm": 0.00029476539930328727, + "kept_token_count": 572.20625, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830308616161346, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.628588598687202e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.4497894287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.66187438964846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6440, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29929390251636506, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.13940241560339928, + "grad_norm": 0.0003223214007448405, + "kept_token_count": 573.2875, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835640370845795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.420319768949412e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.46692810058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3227569580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.30044260025024416, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.14137862920761107, + "grad_norm": 0.0003158716717734933, + "kept_token_count": 572.3921875, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841816604137421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.179054653330241e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.4984558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4031127929687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6460, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.3006018579006195, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.12808551117777825, + "grad_norm": 0.000338624173309654, + "kept_token_count": 572.409375, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9838655591011047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.302536276052706e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.5739807128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.19956665039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6470, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.29947459101676943, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.12870680019259453, + "grad_norm": 0.0003973993589170277, + "kept_token_count": 573.3625, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840694546699524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.222894844540861e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.84747924804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6803680419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6480, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.29889578819274903, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.12527010589838028, + "grad_norm": 0.00032557951635681093, + "kept_token_count": 573.7140625, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839301705360413, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.277294378378428e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.5870086669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.27857360839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6490, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2997030705213547, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.1383529782295227, + "grad_norm": 0.0002717180468607694, + "kept_token_count": 572.8828125, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834767937660217, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.454396461776923e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.7716064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.83766479492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6500, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.29988286197185515, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.13682022392749787, + "grad_norm": 0.00027758299256674945, + "kept_token_count": 572.878125, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9829116761684418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.675141194136813e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.0258087158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75148315429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6510, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2993615299463272, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.13445202261209488, + "grad_norm": 0.0003097579756285995, + "kept_token_count": 573.3359375, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9829684317111969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.65297204250237e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.7089385986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1490020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6520, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30014202296733855, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.1337886244058609, + "grad_norm": 0.00034093507565557957, + "kept_token_count": 572.36875, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831740081310272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.572669262823183e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.1506286621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5563903808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6530, + "valid_token_count": 817.8359375 + }, + { + "actual_token_mask_ratio": 0.30025333166122437, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.13298725262284278, + "grad_norm": 0.00037141816574148834, + "kept_token_count": 572.5734375, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837688028812408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.340335276036057e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.9157196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.56630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6540, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2993023067712784, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.12445028498768806, + "grad_norm": 0.00029435299802571535, + "kept_token_count": 573.0765625, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836822628974915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.374138101818971e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.5466064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5803253173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6550, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.3006139725446701, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.14024016186594962, + "grad_norm": 0.0003410156932659447, + "kept_token_count": 572.3765625, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834870219230651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.450398432207294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.15236511230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2925109863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6560, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29900158643722535, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.1343163289129734, + "grad_norm": 0.00027520646108314395, + "kept_token_count": 573.6515625, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836046099662781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.404469604603946e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.0167266845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7789672851562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6570, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.29909537434577943, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.11170560643076896, + "grad_norm": 0.0003233914030715823, + "kept_token_count": 573.534375, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9821683764457703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.96549268468516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.8038269042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.97014770507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6580, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2994005441665649, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.13886428624391556, + "grad_norm": 0.00025830749655142426, + "kept_token_count": 573.0375, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839227735996247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.280181260081008e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.66026611328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20415649414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6590, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.3005196571350098, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.13684312999248505, + "grad_norm": 0.0003315484500490129, + "kept_token_count": 572.4453125, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9827100932598114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.753888119419571e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.2222534179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7168823242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6600, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2999693602323532, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.13206776157021521, + "grad_norm": 0.000256837229244411, + "kept_token_count": 572.71875, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841369748115539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.196508620632812e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.5458190917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4739654541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.29850887656211855, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.13631733730435372, + "grad_norm": 0.00048068666364997625, + "kept_token_count": 573.9703125, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828481078147888, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.699977748212405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.3144500732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20814819335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6620, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29919248819351196, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.1298379383981228, + "grad_norm": 0.00028219117666594684, + "kept_token_count": 573.4109375, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842150747776032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.166004386614077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.17314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.97633056640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6630, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.29882827401161194, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.13737936019897462, + "grad_norm": 0.0003093216218985617, + "kept_token_count": 574.009375, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836763560771942, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.376439596351701e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.01019897460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0558776855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6640, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.3016610860824585, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.14205725714564324, + "grad_norm": 0.0005180896259844303, + "kept_token_count": 571.4453125, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830194115638733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.633058728766628e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.8710113525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8804046630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6650, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3004367232322693, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.12663782984018326, + "grad_norm": 0.00042845794814638793, + "kept_token_count": 572.3859375, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830344974994659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.627174916502554e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.2580505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6167236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6660, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2997863918542862, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.1363033100962639, + "grad_norm": 0.00039092765655368567, + "kept_token_count": 572.9328125, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833292007446289, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.512056279461831e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.02765502929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42323303222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6670, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3009275496006012, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.1293865256011486, + "grad_norm": 0.00035666298936121166, + "kept_token_count": 571.95, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983563220500946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.420637182600331e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.1173370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.123095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6680, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3002809643745422, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.12406713590025902, + "grad_norm": 0.00033571652602404356, + "kept_token_count": 572.4390625, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841782450675964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.180401396704838e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.5994812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.59070739746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6690, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.29850495755672457, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.13511952906847, + "grad_norm": 0.0002937475801445544, + "kept_token_count": 574.109375, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983183890581131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.568811550096143e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.2588836669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.68292846679685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6700, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29950262010097506, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.14217446967959405, + "grad_norm": 0.00026622266159392893, + "kept_token_count": 573.334375, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832815945148468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.530646387545858e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.80645141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1272308349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6710, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.2992692917585373, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.15108730643987656, + "grad_norm": 0.0002591726079117507, + "kept_token_count": 573.6390625, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834429621696472, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.467608800448943e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.88645629882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7618896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.29924858212471006, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.13706271946430207, + "grad_norm": 0.00029977556550875306, + "kept_token_count": 573.425, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832890093326568, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.527748664666432e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.4820617675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.19679565429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6730, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2994377940893173, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.12446335181593896, + "grad_norm": 0.00027427374152466655, + "kept_token_count": 573.0375, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842233180999755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.162781101011205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.7916259765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.87330017089846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6740, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2999361604452133, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.13557387068867682, + "grad_norm": 0.0003948913363274187, + "kept_token_count": 572.9875, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836069047451019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.403571969713085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.1319244384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.56455688476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6750, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.3009772956371307, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.12949466854333877, + "grad_norm": 0.0003208410053048283, + "kept_token_count": 571.9578125, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839354872703552, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.275216037465725e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.4492248535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.57574768066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6760, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2996042728424072, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.1317862778902054, + "grad_norm": 0.00025190587621182203, + "kept_token_count": 573.10625, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833999574184418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.48440582153853e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.8934326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.27063903808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6770, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.29847516119480133, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.13651594519615173, + "grad_norm": 0.0003488360671326518, + "kept_token_count": 573.8796875, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842429101467133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.155131814011838e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.03626403808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36310424804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6780, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.3003148227930069, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.12955248951911927, + "grad_norm": 0.00033126448397524655, + "kept_token_count": 572.540625, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834934592247009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.44788164208876e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.72296447753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5808044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6790, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.29985072910785676, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.13075660318136215, + "grad_norm": 0.0003277159412391484, + "kept_token_count": 572.8359375, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830217063426971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.632162039750255e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.7903198242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.57371826171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6800, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.30039256513118745, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.1411489687860012, + "grad_norm": 0.00030239243642427027, + "kept_token_count": 572.7046875, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828538715839386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.69772536639357e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.7383056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.81813659667966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6810, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.30027948319911957, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.13730926886200906, + "grad_norm": 0.00046017131535336375, + "kept_token_count": 572.4421875, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9816080570220947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 7.18437491741497e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.83440551757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.39939270019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6820, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.3010715186595917, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.1320688508450985, + "grad_norm": 0.00028970412677153945, + "kept_token_count": 571.7375, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840255081653595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.240058100956958e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.1350067138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2510070800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6830, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.29961802065372467, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.13620625659823418, + "grad_norm": 0.00038867525290697813, + "kept_token_count": 573.175, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.982692950963974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.76058840326732e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.3903350830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6468811035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6840, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.29986631870269775, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.14002587497234345, + "grad_norm": 0.00029729335801675916, + "kept_token_count": 572.8046875, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834813475608826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.45261381578166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.7983001708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6830749511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6850, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.30000036358833315, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.13522567301988603, + "grad_norm": 0.00025910697877407074, + "kept_token_count": 572.8390625, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834291815757752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.472995119111147e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.5881744384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0627197265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6860, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29932997226715086, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.13168401941657065, + "grad_norm": 0.0003438941203057766, + "kept_token_count": 573.3109375, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840186834335327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.242724157345947e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.87705688476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0839599609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6870, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3003416478633881, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.1299478866159916, + "grad_norm": 0.00038561964174732566, + "kept_token_count": 572.5140625, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844589114189148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.070755589462351e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.2392547607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.888427734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6880, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.300617641210556, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.13137951940298082, + "grad_norm": 0.0003443533496465534, + "kept_token_count": 572.146875, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9838040113449097, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.326576294668484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.92799072265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34581604003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6890, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.30094600915908815, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.13403077125549318, + "grad_norm": 0.00030712218722328544, + "kept_token_count": 572.1125, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840495347976684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.230663966562133e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.2864532470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6181671142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6900, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3001646101474762, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.13981341198086739, + "grad_norm": 0.0002754093729890883, + "kept_token_count": 572.553125, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9828034281730652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.717431751894765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.6746032714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.52190551757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6910, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2999452143907547, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.13232600390911103, + "grad_norm": 0.0003862668527290225, + "kept_token_count": 572.725, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842760741710663, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.142172933323308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.8212554931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0558837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6920, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.29926983416080477, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.129766733199358, + "grad_norm": 0.000277079117950052, + "kept_token_count": 573.49375, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835314512252807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.433047055907082e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.1426483154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4856201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6930, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2989661306142807, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.1292843699455261, + "grad_norm": 0.0002725851081777364, + "kept_token_count": 573.7671875, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833617687225342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.499326955236028e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.63330993652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4188537597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6940, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.3008667707443237, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.132436540722847, + "grad_norm": 0.00028993061278015375, + "kept_token_count": 572.1859375, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846190094947815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0082157142460345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.5538665771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.05292053222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6950, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.3003143906593323, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.13739853948354722, + "grad_norm": 0.00034416341804899275, + "kept_token_count": 572.509375, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834916889667511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.448573512898292e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.9800659179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2572296142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6960, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.2986673593521118, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.13499301448464393, + "grad_norm": 0.00019428662199061364, + "kept_token_count": 573.928125, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839569330215454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.266841446631589e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.59021911621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.42267150878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3004094511270523, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.1357491798698902, + "grad_norm": 0.00028837364516220987, + "kept_token_count": 572.6375, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848942816257477, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.90069372265134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.8324005126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.79298095703126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6980, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.29919643700122833, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.12961359992623328, + "grad_norm": 0.00028235255740582943, + "kept_token_count": 573.396875, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840559840202332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.228148304217029e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.3233245849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.09715576171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6990, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29910753965377807, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.12842360585927964, + "grad_norm": 0.00032518411171622574, + "kept_token_count": 573.675, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9831667840480804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.575491170224268e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.1091369628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.16981811523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7000, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.30037021934986113, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.13374029621481895, + "grad_norm": 0.0002938014513347298, + "kept_token_count": 572.371875, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840917706489563, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.214173881744501e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.8304718017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.395166015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7010, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.2997838944196701, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.12518778070807457, + "grad_norm": 0.0002767029218375683, + "kept_token_count": 573.0296875, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842048108577728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.170017913973425e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.8818298339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.58497009277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7020, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.29916744530200956, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.1230512648820877, + "grad_norm": 0.0003461692249402404, + "kept_token_count": 573.5109375, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832476735115051, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.543894087371883e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.1662139892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2029052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7030, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2995936781167984, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.12835180684924125, + "grad_norm": 0.0003143375797662884, + "kept_token_count": 573.053125, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984208208322525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.168682521092705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.8728363037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1540985107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7040, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3004316210746765, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.13030854165554046, + "grad_norm": 0.0003133551508653909, + "kept_token_count": 572.240625, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842938184738159, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.135245384939481e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.5171722412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6027038574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7050, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30014231204986574, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.12567103728652002, + "grad_norm": 0.00031469075474888086, + "kept_token_count": 572.58125, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845054805278778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.052566750440747e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.3268615722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31605224609376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7060, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.299254047870636, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.13080514669418336, + "grad_norm": 0.0003060463641304523, + "kept_token_count": 573.4953125, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9830170631408691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.633983066421934e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.3082305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6893798828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7070, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3003048121929169, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.1257553793489933, + "grad_norm": 0.0002497361565474421, + "kept_token_count": 572.346875, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845302045345307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.042904351488687e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.12781677246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.78106689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30033770501613616, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.1313689984381199, + "grad_norm": 0.00033917257678695023, + "kept_token_count": 572.6359375, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846294164657593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.004155584378168e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.4092559814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0322509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7090, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.29970801472663877, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.13087972700595857, + "grad_norm": 0.00022838891891296953, + "kept_token_count": 572.9484375, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834778666496277, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.453975729527883e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.8326934814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6543762207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7100, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.2997497975826263, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.12773701995611192, + "grad_norm": 0.00039786886190995574, + "kept_token_count": 572.853125, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852559268474579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.759429441241082e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.10072631835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.67088928222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7110, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2998009592294693, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.137123554199934, + "grad_norm": 0.00036710675340145826, + "kept_token_count": 572.8578125, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9838057696819306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.325887407001573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.9888061523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1281372070313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7120, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.30011055171489714, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.14255911186337472, + "grad_norm": 0.00032326183281838894, + "kept_token_count": 572.5921875, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846706569194794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9880485787289216e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.6720855712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.05250854492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7130, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2981570541858673, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.13070151433348656, + "grad_norm": 0.0003125773509964347, + "kept_token_count": 574.21875, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98366978764534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.379005244525616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.57493286132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6259826660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.29990302324295043, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.12504708841443063, + "grad_norm": 0.0002869224699679762, + "kept_token_count": 572.81875, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839407086372376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.273182298173196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.9282684326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1716552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7150, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.3005572706460953, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.1279838725924492, + "grad_norm": 0.00027225431404076517, + "kept_token_count": 572.384375, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9847664833068848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9506115940166636e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.6674011230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.82965698242185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7160, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2993929207324982, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.1297318898141384, + "grad_norm": 0.000334899959852919, + "kept_token_count": 573.4765625, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844656348228454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.06813482590951e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.703076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6778533935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7170, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.30069689452648163, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.12898435145616532, + "grad_norm": 0.00032717661815695465, + "kept_token_count": 572.26875, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844556629657746, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.072032556403428e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.9988098144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.49566650390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7180, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3010521650314331, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.13707703053951265, + "grad_norm": 0.00032432624720968306, + "kept_token_count": 572.2046875, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835408151149749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.429387940443121e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.795458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21771850585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7190, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.30049235224723814, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.1286819323897362, + "grad_norm": 0.0003168389666825533, + "kept_token_count": 572.428125, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846627652645111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9911221978836696e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.7186004638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0566131591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7200, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.29994213581085205, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.1290483959019184, + "grad_norm": 0.0003920057788491249, + "kept_token_count": 572.7578125, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843291699886322, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.121436017565429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.5196258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.95966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7210, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2990644544363022, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.11628646478056907, + "grad_norm": 0.00044493284076452255, + "kept_token_count": 573.7234375, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983966726064682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.263019822654315e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.8534210205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.76694946289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7220, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.29979143738746644, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.1317259632050991, + "grad_norm": 0.0002912454365286976, + "kept_token_count": 572.8484375, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843730807304383, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.10428618529113e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.3032928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3407348632812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7230, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2995861053466797, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.1364610858261585, + "grad_norm": 0.00029111633193679154, + "kept_token_count": 573.1953125, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851218581199646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8117954176850616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.8536834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3060272216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7240, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.298647740483284, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.1299506738781929, + "grad_norm": 0.0003751058829948306, + "kept_token_count": 574.0140625, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985233724117279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7680977988638915e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.9684356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10211791992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7250, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.30061360597610476, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.1293758697807789, + "grad_norm": 0.0003679638321045786, + "kept_token_count": 572.275, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846572935581207, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9932655130978676e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.70953979492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.48663330078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7260, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.3007467180490494, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.140509644895792, + "grad_norm": 0.0003232005692552775, + "kept_token_count": 572.05, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837362229824066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.353052885970101e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.0615692138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1133544921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7270, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.3006050050258636, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.13506472036242484, + "grad_norm": 0.00028300544363446534, + "kept_token_count": 572.36875, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9836903631687164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.370974369929172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.8302459716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24013366699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7280, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.30065602958202364, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.14103584066033364, + "grad_norm": 0.00036417096271179616, + "kept_token_count": 572.1640625, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834231615066529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.475350892287679e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.67625122070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4529113769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7290, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3000116735696793, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.12342845723032951, + "grad_norm": 0.0002617257705423981, + "kept_token_count": 572.740625, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985450291633606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6834988936316225e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.4582458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4687866210937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7300, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29957779943943025, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.13189658969640733, + "grad_norm": 0.00031811234657652676, + "kept_token_count": 573.275, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839697301387786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.261840171646327e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.16961669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.62315979003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7310, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.300249257683754, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.14000898897647857, + "grad_norm": 0.0002609920047689229, + "kept_token_count": 572.5015625, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850371062755585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.84489964239765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.4713928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.32762756347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7320, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.30084115266799927, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.14013196974992753, + "grad_norm": 0.00030358225922100246, + "kept_token_count": 571.834375, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839822888374329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.256932574615348e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.29827880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2781097412109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7330, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.2998756647109985, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.13376881256699563, + "grad_norm": 0.0003353494976181537, + "kept_token_count": 573.03125, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835254609584808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.435386821976863e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.73268737792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8305969238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7340, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.29979380071163175, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.14100813120603561, + "grad_norm": 0.0003376765234861523, + "kept_token_count": 573.071875, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839034259319306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.28774068900384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.7454040527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.81548767089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2999779909849167, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.14364372417330742, + "grad_norm": 0.0003459214058239013, + "kept_token_count": 572.775, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832485377788543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.543563686136621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.0537139892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.58619384765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7360, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29956796765327454, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.13981595188379287, + "grad_norm": 0.00024788721930235624, + "kept_token_count": 573.2, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9837812006473541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.335483230941463e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.1715057373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3116119384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7370, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.30060558319091796, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.13172399401664733, + "grad_norm": 0.00035751753603108227, + "kept_token_count": 572.4, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835877776145935, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.41103779344121e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.0949981689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65887756347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7380, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2986637532711029, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.14443269819021226, + "grad_norm": 0.0002718304458539933, + "kept_token_count": 574.0328125, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846061706542969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.013234742567875e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.83580017089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.02366333007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.29923652708530424, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.12291326522827148, + "grad_norm": 0.0002436725189909339, + "kept_token_count": 573.46875, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841427803039551, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.194245579536073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.2607757568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9674072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7400, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3005728453397751, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.13046210408210754, + "grad_norm": 0.0002747903927229345, + "kept_token_count": 572.2203125, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858173370361328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5401204372174105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.91395568847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5681854248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7410, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3001917898654938, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.12901427671313287, + "grad_norm": 0.0003346970188431442, + "kept_token_count": 572.678125, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852430701255799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.764447305409703e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.5927368164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.64105529785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7420, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30068104565143583, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.12861299142241478, + "grad_norm": 0.0003131419070996344, + "kept_token_count": 572.3734375, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844405710697174, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.077919279050548e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.64703674316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7585876464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7430, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2999582827091217, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.1251490071415901, + "grad_norm": 0.00029718695441260934, + "kept_token_count": 572.778125, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846177339553833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.008713498886209e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.6240936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.5719329833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7440, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2988928765058517, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.12595032081007956, + "grad_norm": 0.0002625652414280921, + "kept_token_count": 573.75, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9835654437541962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.419772362278308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.9580413818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6503112792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7450, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3006439387798309, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.14283453226089476, + "grad_norm": 0.00029964378336444497, + "kept_token_count": 572.025, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849826753139496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8661672301241194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.3464599609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8234619140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7460, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.30027768611907957, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.12928758859634398, + "grad_norm": 0.00022941890347283334, + "kept_token_count": 572.4046875, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852094173431396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7775930326897654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.4191619873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9167816162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7470, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3002789348363876, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.1281077578663826, + "grad_norm": 0.00033029954647645354, + "kept_token_count": 572.7671875, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849141895771026, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8929134320351294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.8993255615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4300140380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7480, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.2999661237001419, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.12454643696546555, + "grad_norm": 0.0002956738171633333, + "kept_token_count": 572.803125, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841536998748779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.189976302266587e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.0516632080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0470306396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7490, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29881471395492554, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.14127480164170264, + "grad_norm": 0.00032586228917352855, + "kept_token_count": 573.759375, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834475517272949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.465823753387667e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.5574066162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.23804626464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7500, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3016136556863785, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.12881219312548636, + "grad_norm": 0.00039500714046880603, + "kept_token_count": 571.4609375, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9832068622112274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.559841640410013e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.01979675292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.69769592285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7510, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3001482516527176, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.13462028354406358, + "grad_norm": 0.0004007399547845125, + "kept_token_count": 572.6390625, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834715247154235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.456455739680677e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.5944549560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2113433837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7520, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.3001990705728531, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.1372762627899647, + "grad_norm": 0.0002706570958252996, + "kept_token_count": 572.45, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846468806266785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.997327825753018e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.90846252441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5020263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.30100990533828736, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.1287834756076336, + "grad_norm": 0.00040181740769185126, + "kept_token_count": 572.178125, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846404075622559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.999857457936741e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.8895721435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0896789550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7540, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.2998524159193039, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.13772772178053855, + "grad_norm": 0.00031518697505816817, + "kept_token_count": 573.2234375, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846454679965972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.997881635266822e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.70216369628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8711456298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7550, + "valid_token_count": 818.71875 + }, + { + "actual_token_mask_ratio": 0.3005128771066666, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.1356290139257908, + "grad_norm": 0.00039206151268444955, + "kept_token_count": 572.3390625, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984312903881073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.1277958957362e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.37994384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9227264404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7560, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.29970181584358213, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.13261476904153824, + "grad_norm": 0.00029200068092904985, + "kept_token_count": 573.1578125, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850850880146027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8261581580154596e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.27073059082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7570, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.2997743308544159, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.13193079680204392, + "grad_norm": 0.00043412623926997185, + "kept_token_count": 572.6125, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840088546276092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.24656764557585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.3678039550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.94993591308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7580, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.30113360583782195, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.12739181220531465, + "grad_norm": 0.00033441183040849864, + "kept_token_count": 571.95625, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850965201854706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8216962133883496e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.7849914550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0816192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7590, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3000319540500641, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.13385220840573311, + "grad_norm": 0.0003087212680839002, + "kept_token_count": 572.7484375, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851969897747039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.782447369711008e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.046435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9503601074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7600, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30011053681373595, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.1335400588810444, + "grad_norm": 0.000296653772238642, + "kept_token_count": 572.75625, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9839881122112274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.254659929254558e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.8830627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.54425659179685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7610, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.29910171329975127, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.13067256286740303, + "grad_norm": 0.0003058284055441618, + "kept_token_count": 573.5859375, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854117870330811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6985384435392913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.6359893798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.04021911621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7620, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.30128723084926606, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.13604215681552886, + "grad_norm": 0.0002821334055624902, + "kept_token_count": 571.7953125, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850099921226502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8554857605486174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.5810241699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3788818359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7630, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.29985267817974093, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.13086448833346367, + "grad_norm": 0.0003269058361183852, + "kept_token_count": 573.203125, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850485980510711, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8404112860444005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.62449951171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8289733886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7640, + "valid_token_count": 818.7 + }, + { + "actual_token_mask_ratio": 0.2992996424436569, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.12500455752015113, + "grad_norm": 0.00036352284951135516, + "kept_token_count": 573.225, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983366709947586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.497397225757595e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.0961090087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.36114501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7650, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.30090650022029874, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.14530357494950294, + "grad_norm": 0.00027731884620152414, + "kept_token_count": 572.140625, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843253314495086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.122931445133873e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.41941528320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8464050292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7660, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2997950166463852, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.13826781287789344, + "grad_norm": 0.00035746960202232003, + "kept_token_count": 573.0625, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845364034175873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.040483640390448e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.22251586914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7192687988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7670, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2994116544723511, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.1331385277211666, + "grad_norm": 0.00026315124705433846, + "kept_token_count": 573.3390625, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857420265674591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5695422997814606e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.9001129150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.15028686523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7680, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29927431643009184, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.12911192774772645, + "grad_norm": 0.00030071428045630455, + "kept_token_count": 573.375, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848521292209625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.917155649513006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.5716857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3994110107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30122403204441073, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.13909280598163604, + "grad_norm": 0.00032787275267764926, + "kept_token_count": 571.678125, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844602108001709, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.070256204111501e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.8653198242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.13233947753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7700, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30002647638320923, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.13450153023004532, + "grad_norm": 0.00031832291278988123, + "kept_token_count": 572.6375, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845959365367889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0172233861521815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.6182403564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4323699951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7710, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2997673362493515, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.12730999961495398, + "grad_norm": 0.0003556807350832969, + "kept_token_count": 573.05, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842895925045013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.136894735391252e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.6169464111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7557434082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7720, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29963639974594114, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.1273467242717743, + "grad_norm": 0.00036363719846121967, + "kept_token_count": 573.04375, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9847014844417572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.975996609777212e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.8561157226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.97279357910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7730, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2996678531169891, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.1294944852590561, + "grad_norm": 0.0002814145409502089, + "kept_token_count": 573.3671875, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848056733608246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9353012693463826e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.1236907958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7337188720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7740, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.2988411456346512, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.1296565927565098, + "grad_norm": 0.00032179919071495533, + "kept_token_count": 573.853125, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849482655525208, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.879605087102391e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.04359436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8008605957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7750, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.3007723420858383, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.13396307379007338, + "grad_norm": 0.0003035297559108585, + "kept_token_count": 572.103125, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983332771062851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.510659477498848e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.86768493652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1176422119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7760, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.30005231499671936, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.13756235390901567, + "grad_norm": 0.0003697802312672138, + "kept_token_count": 572.9328125, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841938555240631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.17429588601226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.322119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2063323974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7770, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.3006707727909088, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.12958745360374452, + "grad_norm": 0.0003038966387975961, + "kept_token_count": 572.3828125, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843748390674592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.103597806941252e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.07773742675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6389434814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3008043527603149, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.13409546092152597, + "grad_norm": 0.00026459299260750413, + "kept_token_count": 572.01875, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850961446762085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.82183671212988e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.28797302246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2023681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7790, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.30015897154808047, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.13704156503081322, + "grad_norm": 0.00030765298288315535, + "kept_token_count": 572.5671875, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851671755313873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.79409948841203e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.923681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3508636474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7800, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.30047631859779356, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.1363137997686863, + "grad_norm": 0.00029035983607172966, + "kept_token_count": 572.503125, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9833968281745911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.48563578579342e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.45411071777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1603607177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7810, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.299820140004158, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.1267114043235779, + "grad_norm": 0.0003371450293343514, + "kept_token_count": 573.0265625, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854157924652099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.696974039892666e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.47980346679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.05409240722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7820, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30091421604156493, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.1332107089459896, + "grad_norm": 0.00024546540225856006, + "kept_token_count": 572.10625, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842871785163879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.137843120086472e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.1646301269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.18287658691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7830, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3009407639503479, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.1330678530037403, + "grad_norm": 0.0003158299659844488, + "kept_token_count": 572.2765625, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985354608297348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.720886983908713e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.75169372558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7637145996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7840, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.29967776238918303, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.12378576174378395, + "grad_norm": 0.0004279521817807108, + "kept_token_count": 573.028125, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843271553516388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.122230224718806e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.2052337646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4403900146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7850, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.3004584550857544, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.13034263476729394, + "grad_norm": 0.0002793304738588631, + "kept_token_count": 572.4015625, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852746248245239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.752116048824973e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.3126983642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.50391540527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.2998924195766449, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.1283671222627163, + "grad_norm": 0.00034634879557415843, + "kept_token_count": 572.771875, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843921780586242, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.096824145060964e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.4208587646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.12225646972655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7870, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.29957795739173887, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.1354747474193573, + "grad_norm": 0.0002794072206597775, + "kept_token_count": 573.05, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851040124893189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.818767640448641e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.2038238525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.68028869628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7880, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.29977058470249174, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.13851578310132026, + "grad_norm": 0.0003202160878572613, + "kept_token_count": 572.8296875, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846217632293701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.00714916799916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.4854736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.26654052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7890, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.3003638505935669, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.13118262514472007, + "grad_norm": 0.00033761002123355865, + "kept_token_count": 572.3765625, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849768161773682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.868452462891583e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.3572052001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5740447998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7900, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29995743930339813, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.1282115623354912, + "grad_norm": 0.00024353066692128778, + "kept_token_count": 572.825, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858063817024231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.54440361156594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.41965942382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6410675048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7910, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2995652198791504, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.13249589055776595, + "grad_norm": 0.0002170820371247828, + "kept_token_count": 573.09375, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855200409889221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.656257053487934e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.01921997070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5820617675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7920, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2997905433177948, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.12795799896121024, + "grad_norm": 0.0002471662010066211, + "kept_token_count": 573.01875, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985703456401825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.584604550676886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.94814147949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.95253295898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7930, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.29941433668136597, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.13654763922095298, + "grad_norm": 0.00030949970823712647, + "kept_token_count": 573.0890625, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846263408660889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.00535211560782e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.0736846923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7160369873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7940, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3011918246746063, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.12915139719843866, + "grad_norm": 0.0002859249652829021, + "kept_token_count": 571.790625, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849979698657989, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.860187375219539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.2002868652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28040771484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3011794060468674, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.12720318138599396, + "grad_norm": 0.00027126603526994586, + "kept_token_count": 571.759375, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985398018360138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.703913666366134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.9196746826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2297882080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7960, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30113540291786195, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.14010992869734765, + "grad_norm": 0.00029802852077409625, + "kept_token_count": 571.81875, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845761001110077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.024982103554066e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.8424011230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.56703491210936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2992099791765213, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.13399112448096276, + "grad_norm": 0.00041048155981115997, + "kept_token_count": 573.3078125, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855303347110749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6522286831750536e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.6125244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0078521728516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7980, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.3001289635896683, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.13926691338419914, + "grad_norm": 0.000361804646672681, + "kept_token_count": 572.7109375, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834761440753936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.4546514840913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.9545379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.29819641113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7990, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.2984730154275894, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.12785274609923364, + "grad_norm": 0.0003105761425103992, + "kept_token_count": 574.0421875, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842047989368439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.170017950353212e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.8285736083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.64414672851564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8000, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2999184250831604, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.12585607245564462, + "grad_norm": 0.0003026035847142339, + "kept_token_count": 572.878125, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852452039718628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7636121709947474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.10455627441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.72474060058596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8010, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3007084935903549, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.12837598323822022, + "grad_norm": 0.00031665837741456926, + "kept_token_count": 572.278125, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857092201709747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.582359081017785e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.79258728027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.10801391601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8020, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.30107050240039823, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.12957035452127458, + "grad_norm": 0.0002859046508092433, + "kept_token_count": 572.0125, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848631680011749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9128462453372777e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.4075469970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1947967529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8030, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.3005092889070511, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.13156660050153732, + "grad_norm": 0.00041372806299477816, + "kept_token_count": 572.646875, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846982061862946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.977280416118447e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.6604370117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7811218261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8040, + "valid_token_count": 818.665625 + }, + { + "actual_token_mask_ratio": 0.29998899400234225, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.13256106078624724, + "grad_norm": 0.0003223280655220151, + "kept_token_count": 572.875, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842827022075653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.13958382018609e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.9906707763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1057952880859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8050, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3005481392145157, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.130362668633461, + "grad_norm": 0.00024261511862277985, + "kept_token_count": 572.38125, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852871656417846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7472204935038464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.71044006347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4776306152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8060, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3003738224506378, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.1377378970384598, + "grad_norm": 0.0002313040749868378, + "kept_token_count": 572.3609375, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851569294929504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7980937708634886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.1092834472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31847229003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8070, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.298808753490448, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.13743822202086448, + "grad_norm": 0.0003082091570831835, + "kept_token_count": 573.84375, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856236696243286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.615771588054486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.6571746826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1556915283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8080, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30090740919113157, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.1380833439528942, + "grad_norm": 0.00030461681308224797, + "kept_token_count": 572.053125, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851893603801727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.785427383671049e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.66868896484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0914642333984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8090, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3000688850879669, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.13690418303012847, + "grad_norm": 0.0003202765074092895, + "kept_token_count": 572.6359375, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9834827840328216, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.452060370065738e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.14515991210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0992767333984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8100, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29896900355815886, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.13493543937802316, + "grad_norm": 0.0003453771641943604, + "kept_token_count": 573.74375, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984808337688446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.934265136602335e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.40497131347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.19228210449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8110, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.3001026690006256, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.13212305158376694, + "grad_norm": 0.0003331830375827849, + "kept_token_count": 572.4265625, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.983702290058136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.3663125547464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.4121856689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3453369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8120, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.2999407023191452, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.13610791340470313, + "grad_norm": 0.00033842973061837256, + "kept_token_count": 572.7265625, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852444648742675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7639035730971955e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.66144409179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8605163574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8130, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29996848702430723, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.13643540814518929, + "grad_norm": 0.0003061529714614153, + "kept_token_count": 572.7140625, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984636515378952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.001378351356834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.98810119628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08599243164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8140, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.299885755777359, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.13546585217118262, + "grad_norm": 0.00033071800135076046, + "kept_token_count": 572.8515625, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986037302017212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.454197926155757e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.1415649414063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.22901611328126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8150, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2997924029827118, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.1268121100962162, + "grad_norm": 0.0003229048161301762, + "kept_token_count": 573.0515625, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844327926635742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0809570277342574e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.46160278320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5631500244141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8160, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30079776644706724, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.13462672531604766, + "grad_norm": 0.00038863232475705445, + "kept_token_count": 572.353125, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845247030258178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.045055088179652e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.79713134765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8111175537109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8170, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.3014276534318924, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.12233916074037551, + "grad_norm": 0.00026324932696297765, + "kept_token_count": 571.3328125, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859067618846893, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.505185035872273e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.63719787597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.0769561767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8180, + "valid_token_count": 817.8578125 + }, + { + "actual_token_mask_ratio": 0.30015060901641843, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.1464965522289276, + "grad_norm": 0.0003360815462656319, + "kept_token_count": 572.6734375, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9840551257133484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.228486163308843e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.0954284667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.08928833007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8190, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.299336251616478, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.12464954853057861, + "grad_norm": 0.0002920081024058163, + "kept_token_count": 573.271875, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855895638465881, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.629092702292837e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.4316650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.11634216308596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2996980458498001, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.12175082862377166, + "grad_norm": 0.00027253958978690207, + "kept_token_count": 572.9203125, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984507954120636, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0516024313983506e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.5122802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.50075073242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8210, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29934987127780915, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.13033813908696174, + "grad_norm": 0.00034482573391869664, + "kept_token_count": 573.3671875, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985533607006073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.650953644362744e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.989892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.58623657226565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3005190849304199, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.1302685782313347, + "grad_norm": 0.00029233063105493784, + "kept_token_count": 572.509375, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850397884845734, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.843848921358585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.0545379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.44432067871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8230, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.3004451423883438, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.13031506687402725, + "grad_norm": 0.0003020131553057581, + "kept_token_count": 572.5078125, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850573718547821, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.83697921683779e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.34021911621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.089111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8240, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.3000772714614868, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.12835844829678536, + "grad_norm": 0.000335666467435658, + "kept_token_count": 572.8, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857385933399201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5708855506964027e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.23675842285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77500610351564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8250, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.29930518567562103, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.13168056607246398, + "grad_norm": 0.0003243840183131397, + "kept_token_count": 573.0390625, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845990061759948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.01603551331209e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.3999816894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.40162353515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8260, + "valid_token_count": 817.8125 + }, + { + "actual_token_mask_ratio": 0.30039175748825075, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.1415110409259796, + "grad_norm": 0.00025887280935421586, + "kept_token_count": 572.56875, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861830174922943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3972725436324256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.3914428710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4887237548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8270, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.299404376745224, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.12792166024446489, + "grad_norm": 0.0003174521552864462, + "kept_token_count": 573.3890625, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855062961578369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6616191068314946e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.5419097900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6247039794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8280, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2990962266921997, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.13664767146110535, + "grad_norm": 0.0003713431069627404, + "kept_token_count": 573.8015625, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842556178569793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.150165500002913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.4662384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0660430908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8290, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.30020169615745546, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.1333196222782135, + "grad_norm": 0.00045340252108871937, + "kept_token_count": 572.65, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846876263618469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9814137784997e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.5881103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5898193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8300, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.30004907846450807, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.13579008281230925, + "grad_norm": 0.0003252538153901696, + "kept_token_count": 572.828125, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857774615287781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.555693533096928e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.4971221923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6158721923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8310, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2990096867084503, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.12970771118998528, + "grad_norm": 0.00037163469824008644, + "kept_token_count": 573.54375, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984492826461792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0575079260161147e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.04823913574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2859344482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8320, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2993734121322632, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.12999791651964188, + "grad_norm": 0.00029849293059669435, + "kept_token_count": 573.06875, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854128956794739, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6981043235282415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.5888641357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2584686279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8330, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.30019554793834685, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.12988277301192283, + "grad_norm": 0.00030245515517890453, + "kept_token_count": 572.5359375, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849668979644776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8723285837913865e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.5024444580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0691711425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8340, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.3001404970884323, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.1284625120460987, + "grad_norm": 0.0002963528095278889, + "kept_token_count": 572.54375, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842251300811767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.162075296742842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.01222534179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7993957519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8350, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.29925637543201444, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.1320376865565777, + "grad_norm": 0.0002452694461680949, + "kept_token_count": 573.4234375, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98548144698143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.67132890864741e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.36122741699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.87824096679685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8360, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.30162228643894196, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.1220533862709999, + "grad_norm": 0.00028358999406918883, + "kept_token_count": 571.24375, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848322629928589, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.924913712078706e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.7609436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.67152709960936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8370, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.30146204233169555, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.13243514075875282, + "grad_norm": 0.00026564786094240844, + "kept_token_count": 571.5734375, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846364259719849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.00141520408215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.9345001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.86542053222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8380, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3001017451286316, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.13310976102948188, + "grad_norm": 0.0002470149483997375, + "kept_token_count": 572.5984375, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856447458267212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6075413158396256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.33042297363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8903015136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8390, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3001094937324524, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.13418232426047325, + "grad_norm": 0.0002452885964885354, + "kept_token_count": 572.6328125, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862583458423615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3678540280088785e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.58935546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3547821044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8400, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.30052213966846464, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.12858822718262672, + "grad_norm": 0.0003950777754653245, + "kept_token_count": 572.3578125, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852635979652404, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.756423925049603e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.02940063476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1435211181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8410, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29928830564022063, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.13144441545009614, + "grad_norm": 0.0003207724366802722, + "kept_token_count": 573.3234375, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852836728096008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.748581388616003e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.0664886474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.27686767578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8420, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.30027703642845155, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.1313871718943119, + "grad_norm": 0.00022184009139891714, + "kept_token_count": 572.6984375, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852895200252533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.746305432694498e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.31104736328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2819854736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8430, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.30144464671611787, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.13965441957116126, + "grad_norm": 0.0002920401166193187, + "kept_token_count": 571.6359375, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851521909236908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.799942337034736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.131005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.18796691894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8440, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3006273955106735, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.1397462949156761, + "grad_norm": 0.0002649505331646651, + "kept_token_count": 572.46875, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854802548885345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.671792860084679e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.3302062988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1853088378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8450, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.2990525811910629, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.13235675394535065, + "grad_norm": 0.00027334457263350487, + "kept_token_count": 573.6109375, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855381190776825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6491922441637145e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.3729766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3204071044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8460, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2998789846897125, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.13637519180774688, + "grad_norm": 0.00030397201771847904, + "kept_token_count": 572.64375, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858679234981537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.520363556570374e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.489208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5290496826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8470, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.3010595142841339, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.13257612213492392, + "grad_norm": 0.0002833823091350496, + "kept_token_count": 571.98125, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859530389308929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.487114322022535e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.8235107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0101348876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8480, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2992301046848297, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.12928312197327613, + "grad_norm": 0.0003391047357581556, + "kept_token_count": 573.2640625, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851135194301606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.815052718389779e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.9305145263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9496246337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8490, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.29992154240608215, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.13510102033615112, + "grad_norm": 0.000391482375562191, + "kept_token_count": 572.725, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850926756858825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.823193423566408e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.5001525878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9880401611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8500, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3007815808057785, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.12643133625388145, + "grad_norm": 0.0004204873985145241, + "kept_token_count": 572.08125, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857652604579925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5604645604034884e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.51804504394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.83213195800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8510, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30040630102157595, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.1302443191409111, + "grad_norm": 0.00025108858244493604, + "kept_token_count": 572.5203125, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850311696529388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8472128148423505e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.7720001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3694702148438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8520, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29906874895095825, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.1363597497344017, + "grad_norm": 0.0002603490138426423, + "kept_token_count": 573.646875, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859165728092194, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.501357009052299e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.2610687255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4700408935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8530, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.299888676404953, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.14092871472239493, + "grad_norm": 0.0002471590123604983, + "kept_token_count": 572.6703125, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985579913854599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6328603750444015e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.56464233398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.36101989746095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8540, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.30087246298789977, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.12479835078120231, + "grad_norm": 0.0002847201540134847, + "kept_token_count": 572.1953125, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859250307083129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.498054415511433e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.53892517089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.48174438476565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8550, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3007330119609833, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.14035573229193687, + "grad_norm": 0.0002968805783893913, + "kept_token_count": 572.1421875, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849285185337067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8873173111351206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.50278930664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5336669921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8560, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30052519738674166, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.14115894958376884, + "grad_norm": 0.0002629264781717211, + "kept_token_count": 572.4328125, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9844369769096375, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.07932175626047e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.4011535644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6070129394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8570, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2994570881128311, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.13448796197772026, + "grad_norm": 0.0003338445967528969, + "kept_token_count": 573.3859375, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860480546951294, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4499992984347045e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.74139709472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8578216552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8580, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3002798676490784, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.12597687616944314, + "grad_norm": 0.00029062555404379964, + "kept_token_count": 572.6109375, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853872835636139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.708112621505279e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.62401428222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1114990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30066968202590943, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.13613219559192657, + "grad_norm": 0.0003608891856856644, + "kept_token_count": 572.4078125, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850470304489136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.841021047672257e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.79078063964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.93211364746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8600, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.29974419474601743, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.13032421916723252, + "grad_norm": 0.0002874091442208737, + "kept_token_count": 573.234375, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848064303398132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.93500863033114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.5697845458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.247705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8610, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.299074649810791, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.1331000953912735, + "grad_norm": 0.00026356102898716927, + "kept_token_count": 573.5484375, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845145344734192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.0490306714200416e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.59111938476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2518737792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8620, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3003339320421219, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.1302999250590801, + "grad_norm": 0.00022580022050533444, + "kept_token_count": 572.528125, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861446380615234, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4122630172059874e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.14686889648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3863983154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8630, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3005137324333191, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.13342595025897025, + "grad_norm": 0.00028241905965842307, + "kept_token_count": 572.3953125, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856614649295807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.60100579605205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.32066955566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7115905761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8640, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3003698319196701, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.13748320788145066, + "grad_norm": 0.00025963151711039245, + "kept_token_count": 572.396875, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860156178474426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.462666304083541e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.74030151367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4950805664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29987206161022184, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.12766711488366128, + "grad_norm": 0.00035704506444744766, + "kept_token_count": 572.85, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856239080429077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.615671143459622e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.00665893554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.60003356933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8660, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30028864145278933, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.14004875719547272, + "grad_norm": 0.00037656290805898607, + "kept_token_count": 572.4296875, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.984476524591446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.063877008273266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.0800384521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1168548583984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8670, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2995469242334366, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.1421181820333004, + "grad_norm": 0.0003522769547998905, + "kept_token_count": 573.1015625, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852820575237274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.749214069510345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.21270751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0070404052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8680, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29976445734500884, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.12627899646759033, + "grad_norm": 0.000339354359311983, + "kept_token_count": 572.8046875, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854760348796845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.673437444784213e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.1592590332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.20753784179686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8690, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.2994931012392044, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.12132881060242653, + "grad_norm": 0.00029277894645929337, + "kept_token_count": 573.253125, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853371679782867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.727688294427935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.1684997558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.16253662109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29978629350662234, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.14036798030138015, + "grad_norm": 0.0003335139190312475, + "kept_token_count": 572.8578125, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9841213703155518, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.202608492458239e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.50379943847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.35137634277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8710, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29913129806518557, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.13399065658450127, + "grad_norm": 0.0002588435309007764, + "kept_token_count": 573.640625, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848776519298553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.907191516598687e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.51466064453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24268798828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8720, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.29779503047466277, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.1297355853021145, + "grad_norm": 0.0003298948868177831, + "kept_token_count": 574.9203125, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857779920101166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.555491807172075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.52439270019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1127227783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8730, + "valid_token_count": 818.7296875 + }, + { + "actual_token_mask_ratio": 0.2989785701036453, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.1337301179766655, + "grad_norm": 0.0002657084842212498, + "kept_token_count": 573.6453125, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986265230178833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.365160213841591e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.9519775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.33509216308596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8740, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2994493246078491, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.1273296743631363, + "grad_norm": 0.0002798586501739919, + "kept_token_count": 573.128125, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9845465183258056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.036536615283694e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.10483703613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2101654052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8750, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.3001777768135071, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.1283758871257305, + "grad_norm": 0.00027358808438293636, + "kept_token_count": 572.8265625, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854203045368195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.695211912097875e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.86444702148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6237548828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8760, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3004640132188797, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.14477957412600517, + "grad_norm": 0.0003642476222012192, + "kept_token_count": 572.278125, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858152866363525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5409130800399e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.1248718261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7938232421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8770, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.2993254780769348, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.13525475114583968, + "grad_norm": 0.00031310791382566094, + "kept_token_count": 573.459375, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852907001972199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7458372248220256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.2523712158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8838836669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8780, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.29933195412158964, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.13120756149291993, + "grad_norm": 0.0002866554423235357, + "kept_token_count": 573.325, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849127173423767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.89349030633457e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.7771728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.97608642578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8790, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.3006630152463913, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.12736710608005525, + "grad_norm": 0.00030803325353190303, + "kept_token_count": 572.30625, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852295160293579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.769745803263504e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.9703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3181488037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3005663424730301, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.1374252773821354, + "grad_norm": 0.0002453781198710203, + "kept_token_count": 572.4109375, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9847790777683259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.945686461927835e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.9390441894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6940551757813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8810, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.29957049489021303, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.14579362124204637, + "grad_norm": 0.0002217896399088204, + "kept_token_count": 573.19375, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985418725013733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6958245841087775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6119079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9653289794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8820, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3000853270292282, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.13462733700871468, + "grad_norm": 0.0002317382168257609, + "kept_token_count": 572.8375, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98531693816185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.735585946240462e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.2100524902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.57166748046876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2994969367980957, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.13533584773540497, + "grad_norm": 0.00022944055672269315, + "kept_token_count": 573.1984375, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857939183712006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5492698447778824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.3842498779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6550628662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8840, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2998588055372238, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.13607240915298463, + "grad_norm": 0.00024159062013495713, + "kept_token_count": 572.86875, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852143883705139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.775642930530012e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.04628295898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.31682739257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8850, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3002632647752762, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.13117026910185814, + "grad_norm": 0.00028117321198806167, + "kept_token_count": 572.590625, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858125209808349, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.541997852560598e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.83213500976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6029479980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8860, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2990275859832764, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.13644563853740693, + "grad_norm": 0.00032562139676883817, + "kept_token_count": 573.55, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9847213566303253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9682408027583735e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.52304992675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61867370605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8870, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2994591861963272, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.1256827361881733, + "grad_norm": 0.00028995645698159933, + "kept_token_count": 573.3046875, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985391628742218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.706412339350209e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.5235290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4785491943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2995742231607437, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.13248736634850503, + "grad_norm": 0.00023484705889131874, + "kept_token_count": 573.2, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860235810279846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.459553321998101e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.6342468261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.48414611816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8890, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30028088986873624, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.12881915047764778, + "grad_norm": 0.0002401151868980378, + "kept_token_count": 572.5796875, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855944037437439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.627204118354712e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.98553771972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3406005859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8900, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29958424866199496, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.13087298795580865, + "grad_norm": 0.00026554649230092764, + "kept_token_count": 573.2203125, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859452843666077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.490139119501691e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.7016906738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9774444580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2994271367788315, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.13163652941584586, + "grad_norm": 0.00028129987185820937, + "kept_token_count": 573.109375, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850878119468689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8250950678484516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.71819458007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8920, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.30062214732170106, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.1422320105135441, + "grad_norm": 0.00042203214252367616, + "kept_token_count": 572.2375, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9846063673496246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.013154379616026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.84921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.38628845214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8930, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29986668527126314, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.13927317708730697, + "grad_norm": 0.00042994352406822145, + "kept_token_count": 573.046875, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985724663734436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5763209456927146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.0133361816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.22767639160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8940, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.29956514835357667, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.136679108440876, + "grad_norm": 0.00026942946715280414, + "kept_token_count": 573.35625, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858824074268341, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.514703116205055e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.77789001464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4340118408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8950, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.30139686465263366, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.13897961005568504, + "grad_norm": 0.00032538746017962694, + "kept_token_count": 571.5546875, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857329428195953, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5730856183799914e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.3470489501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8734985351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8960, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.30072581171989443, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.1331304520368576, + "grad_norm": 0.00024399123503826559, + "kept_token_count": 572.0890625, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859837055206299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.475127909448929e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.8984680175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36110229492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8970, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.29999264478683474, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.13092206716537474, + "grad_norm": 0.0002858962689060718, + "kept_token_count": 572.6796875, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849898099899292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.86337464483222e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.8972869873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8391937255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8980, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.30052203238010405, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.12846817076206207, + "grad_norm": 0.00021386136359069496, + "kept_token_count": 572.428125, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854588985443116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.680140347976703e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.04075317382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7171569824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8990, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.30014083385467527, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.12533882781863212, + "grad_norm": 0.00023135558876674622, + "kept_token_count": 572.60625, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862537145614624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3696587565355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.4234588623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0244567871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9000, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.3007660061120987, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.13869589269161225, + "grad_norm": 0.00037926604272797704, + "kept_token_count": 572.375, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854082584381103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.699914145225193e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.09898376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1612030029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9010, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.3008542269468307, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.1220555104315281, + "grad_norm": 0.0004129047447349876, + "kept_token_count": 572.0609375, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853299379348754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.730510711146053e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.03069458007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8584411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9020, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3004463344812393, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.1329788640141487, + "grad_norm": 0.0003102085320279002, + "kept_token_count": 572.465625, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9842673718929291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.145580518932548e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.9459259033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5378845214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9030, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.2997794449329376, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.13182006701827048, + "grad_norm": 0.000233235172345303, + "kept_token_count": 572.9390625, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848665773868561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.911518383072689e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.93409423828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.14821166992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9040, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3004226118326187, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.12515992000699044, + "grad_norm": 0.0002703821810428053, + "kept_token_count": 572.70625, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9847666680812835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.950537706667092e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.45701293945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7423431396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9050, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.30088270008563994, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.1392396979033947, + "grad_norm": 0.0003037644491996616, + "kept_token_count": 572.275, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856752276420593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.595637667283881e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.3862823486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.69093933105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9060, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.29949085116386415, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.1308617778122425, + "grad_norm": 0.00027687475085258484, + "kept_token_count": 573.203125, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849055171012878, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.896293332625646e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.4894958496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3349975585937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9070, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30056850612163544, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.12768795639276503, + "grad_norm": 0.00028032067348249257, + "kept_token_count": 572.5765625, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857208371162415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5778150272089985e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.05670166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.12395629882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9080, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.30121639668941497, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.12767357528209686, + "grad_norm": 0.0004426349187269807, + "kept_token_count": 571.9296875, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853451311588287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7245727293775414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.54095458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.65607299804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9090, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.29979204535484316, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.12863779217004775, + "grad_norm": 0.00026378838811069727, + "kept_token_count": 572.98125, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855695843696595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6368953300989234e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.44268493652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.86679992675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9100, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.30046252012252805, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.12334510460495948, + "grad_norm": 0.00027185396174900234, + "kept_token_count": 572.453125, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854124546051025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6982861497090195e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3241882324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6695617675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3001429051160812, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.1281704418361187, + "grad_norm": 0.0002856669598259032, + "kept_token_count": 572.6625, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854619562625885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.678943234670442e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.1765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1063201904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9120, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29993515014648436, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.12678357437253, + "grad_norm": 0.00021495124383363873, + "kept_token_count": 572.9453125, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986722332239151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.186607522773556e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.9335540771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77308044433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9130, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2995942264795303, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.1295054465532303, + "grad_norm": 0.0003124391951132566, + "kept_token_count": 573.2046875, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851422369480133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8038372662849726e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.24231872558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4793243408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9140, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29851376116275785, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.1410634122788906, + "grad_norm": 0.0002905365836340934, + "kept_token_count": 574.0734375, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849356949329376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.884519196115434e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.1112426757812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5337829589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9150, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.3000264227390289, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.12545961737632752, + "grad_norm": 0.00027887403848581016, + "kept_token_count": 572.7296875, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851894736289978, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.785381326859351e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.77252502441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4305877685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9160, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.30058546364307404, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.13197204917669297, + "grad_norm": 0.00028927429229952395, + "kept_token_count": 572.2875, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854282796382904, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.692097147402819e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.28467712402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9755798339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9170, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2998821049928665, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.12693374529480933, + "grad_norm": 0.00023744312056805938, + "kept_token_count": 572.684375, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852798044681549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.750088857894298e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.5276672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0253204345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9180, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.30065867900848386, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.13712436631321906, + "grad_norm": 0.0002703342470340431, + "kept_token_count": 572.3640625, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855170249938965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.657434303429909e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.7123199462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.32508850097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9190, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29921933114528654, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.13243754729628562, + "grad_norm": 0.0003322184784337878, + "kept_token_count": 573.365625, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858674526214599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.520547201740555e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.6287322998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9404357910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3004468411207199, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.13464040011167527, + "grad_norm": 0.00043819111306220293, + "kept_token_count": 572.309375, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857329964637757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.573068774538115e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.74830017089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.65342407226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9210, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.3008186936378479, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.1374307319521904, + "grad_norm": 0.0002891802869271487, + "kept_token_count": 572.0703125, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850244104862214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.849852750543505e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.7120697021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7003875732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9220, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.30103358924388884, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.1353330485522747, + "grad_norm": 0.0002391871967120096, + "kept_token_count": 572.0765625, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9843203842639923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 6.124868014012463e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.6774475097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.15455932617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9230, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.300543349981308, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.12134196311235428, + "grad_norm": 0.0003719569940585643, + "kept_token_count": 572.403125, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858494639396668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.527575849555433e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.58318786621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4873504638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9240, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.3004147559404373, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.12296601086854934, + "grad_norm": 0.00024662064970470965, + "kept_token_count": 572.5125, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859748244285583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.478600178321358e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.68918762207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2650573730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9250, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3011907935142517, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.13098069652915, + "grad_norm": 0.0002671518886927515, + "kept_token_count": 571.7328125, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859073877334594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.504947548615746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.02911071777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5647003173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9260, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.2996948003768921, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.13084639459848404, + "grad_norm": 0.00027685612440109253, + "kept_token_count": 573.384375, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860402584075928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.453040685097221e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.82838439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.02293701171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9270, + "valid_token_count": 818.765625 + }, + { + "actual_token_mask_ratio": 0.30039126574993136, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.12886624410748482, + "grad_norm": 0.0003202750813215971, + "kept_token_count": 572.6375, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858535945415496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.525951455638278e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.7274566650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9475799560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9280, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.300000986456871, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.12564342990517616, + "grad_norm": 0.00028191370074637234, + "kept_token_count": 572.7078125, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869311809539795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.105024865770247e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.8546997070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9290, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2989289253950119, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.1354428067803383, + "grad_norm": 0.0003243717656005174, + "kept_token_count": 573.590625, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864148736000061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.306707025738433e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.7423614501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.25938720703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9300, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.29984220266342165, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.13152143508195877, + "grad_norm": 0.0002459690149407834, + "kept_token_count": 572.6875, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857250869274139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.576155199378263e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.7499664306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.37823486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9310, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.2996801048517227, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.12499770671129226, + "grad_norm": 0.0002646352513693273, + "kept_token_count": 573.2125, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857963263988495, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.548327535507269e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.880126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.5569213867187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9320, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.29932601749897003, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.1281588666141033, + "grad_norm": 0.00030731636798009276, + "kept_token_count": 573.3734375, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852150917053223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.775373756478075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.5489196777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8313415527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9330, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.300367659330368, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.13797125816345215, + "grad_norm": 0.0002581053995527327, + "kept_token_count": 572.5390625, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858267366886139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5364492436638105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.7350158691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.73602294921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9340, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3016157031059265, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.13653429746627807, + "grad_norm": 0.00030067458283156157, + "kept_token_count": 571.50625, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855195164680481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.65645608730847e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.75482177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71303405761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.299673929810524, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.1247285321354866, + "grad_norm": 0.0004411644476931542, + "kept_token_count": 573.1265625, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98590087890625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.507493078766856e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.01001892089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.99562072753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3001999706029892, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.11926739662885666, + "grad_norm": 0.0002882217231672257, + "kept_token_count": 572.6203125, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855321407318115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6515263713663445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.0794250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10472412109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2993916928768158, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.12282330840826035, + "grad_norm": 0.0002781795337796211, + "kept_token_count": 573.2609375, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864751636981964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.283157261146698e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.4403350830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6857025146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30028448104858396, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.1250121735036373, + "grad_norm": 0.0003151012642774731, + "kept_token_count": 572.696875, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854131698608398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.697994056390599e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.7034210205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6665496826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9390, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.30000067949295045, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.1282130792737007, + "grad_norm": 0.0003199272614438087, + "kept_token_count": 572.821875, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854787111282348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6723957823123786e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.28094482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9400, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2994656503200531, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.13464301377534865, + "grad_norm": 0.0003039152070414275, + "kept_token_count": 573.3921875, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9849688529968261, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.871557914360892e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.76588439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7230560302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9410, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.2994322896003723, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.13061175420880317, + "grad_norm": 0.00030020985286682844, + "kept_token_count": 573.2875, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863732993602753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.322950601112097e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.6205627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8073669433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9420, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.29957679808139803, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.12913156673312187, + "grad_norm": 0.0003844180318992585, + "kept_token_count": 573.453125, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864923655986786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2764412976102905e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.9637847900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8102783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9430, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.300110799074173, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.13469610139727592, + "grad_norm": 0.0002811966696754098, + "kept_token_count": 572.7171875, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986021202802658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.460481115733273e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.4199554443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1416046142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9440, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.3002056062221527, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.13473947197198868, + "grad_norm": 0.000307051173876971, + "kept_token_count": 572.4265625, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862917542457581, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3547978677670474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.88812866210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.247216796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9450, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.3000137388706207, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.12654769122600557, + "grad_norm": 0.00035150270559825003, + "kept_token_count": 572.71875, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9848816156387329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.9056361351395026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.1910858154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8279571533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9460, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.30052670538425447, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.13951053619384765, + "grad_norm": 0.00026726245414465666, + "kept_token_count": 572.5484375, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853778600692749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.711798257834744e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.88414306640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6199981689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9470, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29942916333675385, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.1280273087322712, + "grad_norm": 0.0002552298246882856, + "kept_token_count": 573.175, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858319282531738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5344203428830954e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.07093200683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7255462646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9480, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30004067718982697, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.1357313558459282, + "grad_norm": 0.000256012543104589, + "kept_token_count": 572.628125, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861426472663879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.41304936632514e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.92332153320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1210906982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9490, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.3002177864313126, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.11795024573802948, + "grad_norm": 0.00026535612414591014, + "kept_token_count": 572.5859375, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857183158397674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5787977908039464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.499755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3066772460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9500, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30114566087722777, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.13165267184376717, + "grad_norm": 0.00024789603776298463, + "kept_token_count": 571.9125, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867903053760528, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.160057698958553e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.9797027587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.44744873046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9510, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29908147752285, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.1381424717605114, + "grad_norm": 0.0003098642046097666, + "kept_token_count": 573.2328125, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854202151298523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6952474187710324e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.1833770751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.39117126464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9520, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.30098555982112885, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.12313127145171165, + "grad_norm": 0.000358887278707698, + "kept_token_count": 571.98125, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855069041252136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.661383620463312e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.5089416503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.01630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9530, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2994871109724045, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.12982903718948363, + "grad_norm": 0.0002472074411343783, + "kept_token_count": 573.284375, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862658739089966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.364913267840166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.7396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0933410644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9540, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29909362494945524, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.1264277033507824, + "grad_norm": 0.00020627873891498893, + "kept_token_count": 573.615625, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867295861244202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.183773428143468e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.32408752441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.38995666503905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9550, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.2991574019193649, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.12174748331308365, + "grad_norm": 0.0002815975749399513, + "kept_token_count": 573.5546875, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859987795352936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.469246862048749e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.5281127929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.31268005371095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9560, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.29956393241882323, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.12710409387946128, + "grad_norm": 0.00034973525907844305, + "kept_token_count": 572.9828125, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862895667552948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.355647808755748e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.977392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8732604980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9570, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.3010323256254196, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.1323707014322281, + "grad_norm": 0.00024815284996293485, + "kept_token_count": 572.2, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859988272190094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.469227471621707e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.7014221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7727783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9580, + "valid_token_count": 818.64375 + }, + { + "actual_token_mask_ratio": 0.3006531983613968, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.12619582787156106, + "grad_norm": 0.0003067988436669111, + "kept_token_count": 572.1828125, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852768778800964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.751237222284544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.94007873535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.02879638671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.3009162753820419, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.12800973728299142, + "grad_norm": 0.0003395242092665285, + "kept_token_count": 572.15, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863964438438415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.313907204254065e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.340673828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7155822753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9600, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2994381755590439, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.13464832529425622, + "grad_norm": 0.0002337910991627723, + "kept_token_count": 573.0953125, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860917150974273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.432950893009547e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.80869750976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.33814697265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9610, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.29945560693740847, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.13324380442500114, + "grad_norm": 0.00030433511710725725, + "kept_token_count": 573.21875, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853680491447449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.715623847208917e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.58970947265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10931701660155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9620, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29908611476421354, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.1287979006767273, + "grad_norm": 0.00028091782587580383, + "kept_token_count": 573.5171875, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986769825220108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.168058669369202e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.81189880371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0616760253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9630, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30000804662704467, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.13139892369508743, + "grad_norm": 0.00025547773111611605, + "kept_token_count": 572.6953125, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865812599658966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.241714607109316e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.0780975341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4434051513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9640, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.30062311291694643, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.134080208837986, + "grad_norm": 0.00024823949206620455, + "kept_token_count": 572.3828125, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856746554374695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5958582015591675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.1851379394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8863891601562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9650, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.29960538148880006, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.12263929471373558, + "grad_norm": 0.0002488325990270823, + "kept_token_count": 573.325, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869351983070374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.103455332573503e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.8524932861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0708831787109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9660, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.3002990663051605, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.12203975915908813, + "grad_norm": 0.00032292521791532636, + "kept_token_count": 572.5859375, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861759066581726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.400057561928406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.3275451660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4269958496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3004222124814987, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.12356669306755066, + "grad_norm": 0.00022565409017261118, + "kept_token_count": 572.4515625, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870417296886445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.061846604803577e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.6055480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2839111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9680, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30007700622081757, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.12971486449241637, + "grad_norm": 0.000301074207527563, + "kept_token_count": 572.6015625, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865970373153686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.235554199316539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.35198974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7655731201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9690, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3003288745880127, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.1344605952501297, + "grad_norm": 0.0003207608824595809, + "kept_token_count": 572.534375, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985650098323822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.605445403489284e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.4481689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2281860351562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9700, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2984155654907227, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.12697651982307434, + "grad_norm": 0.00029705496854148805, + "kept_token_count": 574.1015625, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859125316143036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.502936655830126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.13629455566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5490997314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9710, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.3004998445510864, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.13015830516815186, + "grad_norm": 0.00023404703824780881, + "kept_token_count": 572.49375, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864789128303528, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.281693411234301e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.1152099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.78789672851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9720, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29966997504234316, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.11818621903657914, + "grad_norm": 0.00030301688821054995, + "kept_token_count": 573.09375, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864188134670258, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.305171544023324e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.46197814941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7375518798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9730, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30101247429847716, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.1341065473854542, + "grad_norm": 0.00035685181501321495, + "kept_token_count": 571.83125, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856705665588379, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.597453819063958e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.2630218505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.71990661621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9740, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.30060801804065707, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.1249579906463623, + "grad_norm": 0.00025676912628114223, + "kept_token_count": 572.303125, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858866393566131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.513056203199085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.50941467285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3224243164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9750, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30008792877197266, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.1351414643228054, + "grad_norm": 0.00029125006403774023, + "kept_token_count": 572.5953125, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856307029724121, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.613026150967926e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.1159240722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.14589233398436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9760, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29897571802139283, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.1215412050485611, + "grad_norm": 0.00024601389304734766, + "kept_token_count": 573.7953125, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859090864658355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.504284927155822e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.27028198242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.58318786621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9770, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.29970278441905973, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.13080323934555055, + "grad_norm": 0.0002774735912680626, + "kept_token_count": 572.8515625, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857334315776825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.572895461227745e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.5834930419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5116790771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9780, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.3000939577817917, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.13402116000652314, + "grad_norm": 0.00039336265763267875, + "kept_token_count": 572.8, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852926194667816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7450887106824665e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.3714660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8852996826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9790, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.3007765978574753, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.1190901406109333, + "grad_norm": 0.0002690812398213893, + "kept_token_count": 572.31875, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866985559463501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1958900439785796e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.8688110351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.764501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9800, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.2997062742710114, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.12537832483649253, + "grad_norm": 0.00032475529587827623, + "kept_token_count": 572.821875, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853979349136353, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7039492821786554e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.6367950439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4997589111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9810, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.2998459905385971, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.13089170232415198, + "grad_norm": 0.00025921969790942967, + "kept_token_count": 572.909375, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850311875343323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8472102682571855e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.82090454101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3090362548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9820, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30007914900779725, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.13447118327021598, + "grad_norm": 0.00028028269298374653, + "kept_token_count": 572.74375, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869786143302918, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.086499368189834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.9349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.18548278808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2991537362337112, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.13105012774467467, + "grad_norm": 0.00022526219254359603, + "kept_token_count": 573.3375, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859914124011994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.472123630170245e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.20733337402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75450744628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9840, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.3006666272878647, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.1300386905670166, + "grad_norm": 0.0004131144087295979, + "kept_token_count": 571.9453125, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862685084342957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.363881318771746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.3305419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6316223144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9850, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.2994392096996307, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.12570186331868172, + "grad_norm": 0.0003190264105796814, + "kept_token_count": 573.040625, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864961326122283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.274964969430585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.9738006591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4907653808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9860, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.30066058933734896, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.13641244694590568, + "grad_norm": 0.0002208910445915535, + "kept_token_count": 572.61875, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865870952606202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.239439342403784e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.93531799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2748199462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9870, + "valid_token_count": 818.803125 + }, + { + "actual_token_mask_ratio": 0.3003878712654114, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.13135884404182435, + "grad_norm": 0.0003366700548212975, + "kept_token_count": 572.4625, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859620332717896, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.483600034494884e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.2721313476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.06935424804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9880, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29987357556819916, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.13543718829751014, + "grad_norm": 0.0003552610578481108, + "kept_token_count": 572.9203125, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858801007270813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5156012604129503e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.69097595214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8685546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9890, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2996490806341171, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.13008812367916106, + "grad_norm": 0.0002769329585134983, + "kept_token_count": 573.0296875, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986671245098114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.206562600506004e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.822314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.07232055664065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9900, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30110050439834596, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.127960404753685, + "grad_norm": 0.0002615097037050873, + "kept_token_count": 572.0796875, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864931106567383, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2761498591280545e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.6881469726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5231903076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9910, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.29939640164375303, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.13380674198269843, + "grad_norm": 0.000236572785070166, + "kept_token_count": 573.14375, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862379729747772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3758085050503725e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.88604736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8722320556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9920, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29939873814582824, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.13903606832027435, + "grad_norm": 0.0003456641570664942, + "kept_token_count": 573.315625, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861578464508056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.407114585977979e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.4665069580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9298431396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9930, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29975921511650083, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.12799490690231324, + "grad_norm": 0.0002785974647849798, + "kept_token_count": 573.1078125, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869243383407593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.107695142214652e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.3683624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5764587402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9940, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.29950627088546755, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.1284836895763874, + "grad_norm": 0.00026082887779921293, + "kept_token_count": 573.0640625, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867219924926758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.186742600926664e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.8913970947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.64013671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30073025822639465, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.127104938775301, + "grad_norm": 0.0002774089516606182, + "kept_token_count": 572.121875, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860134482383728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4635171181871556e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.66075134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.808837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9960, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.300078472495079, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.1328807145357132, + "grad_norm": 0.0002612248354125768, + "kept_token_count": 572.9, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9854149639606475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.69729701965116e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.35695495605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.72132263183596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9970, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.30092655420303344, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.12215347662568092, + "grad_norm": 0.00022392913524527103, + "kept_token_count": 572.26875, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862334311008454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.377583875088021e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.519775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3125396728516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9980, + "valid_token_count": 818.603125 + }, + { + "actual_token_mask_ratio": 0.2992384672164917, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.12866942584514618, + "grad_norm": 0.0002490236656740308, + "kept_token_count": 573.409375, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865121603012085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.268703680485487e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.6609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.73150024414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9990, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3002124935388565, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.13075325936079024, + "grad_norm": 0.0002594284887891263, + "kept_token_count": 572.5125, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862751126289367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.36130817636149e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.46697387695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.71885375976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10000, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.29941655695438385, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.12462249472737312, + "grad_norm": 0.00037729842006228864, + "kept_token_count": 573.434375, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861585199832916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4068506869953126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.99432678222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9934814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10010, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.3006997436285019, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.13135272562503814, + "grad_norm": 0.0002863768895622343, + "kept_token_count": 572.265625, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866966366767883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1966456885566005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.2992797851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5078948974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10020, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3002713471651077, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.12695711553096772, + "grad_norm": 0.00026378626353107393, + "kept_token_count": 572.828125, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865548074245453, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.25204526638845e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.2091552734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.20517578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10030, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.30035417079925536, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.12264924347400666, + "grad_norm": 0.0003155973390676081, + "kept_token_count": 572.4296875, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866676092147827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.207980611885432e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.61155090332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.43804931640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2991264373064041, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.13214579820632935, + "grad_norm": 0.0003142689529340714, + "kept_token_count": 573.4078125, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866695404052734, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.207226895436179e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.1239837646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1666625976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2995853126049042, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.1207790955901146, + "grad_norm": 0.0003080852038692683, + "kept_token_count": 573.178125, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861022770404816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4288161845761354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.9823272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8152221679687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10060, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29984872341156005, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.13276151418685914, + "grad_norm": 0.0002688108943402767, + "kept_token_count": 572.9578125, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857459962368011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.567993794102222e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.6206390380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4584594726563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10070, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3006237745285034, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.13150811344385147, + "grad_norm": 0.0002776798792183399, + "kept_token_count": 572.09375, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860538959503173, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.447712137538474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.4714111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0230987548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10080, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.29965250194072723, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.12200710326433181, + "grad_norm": 0.00031173485331237316, + "kept_token_count": 573.015625, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859910547733307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4722649656469004e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.7345031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6785400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10090, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.300640469789505, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.1361183226108551, + "grad_norm": 0.0003098729357589036, + "kept_token_count": 572.359375, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986972987651825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.088700017950032e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.8943115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5255401611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3014000028371811, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.13836959525942802, + "grad_norm": 0.00034067381056956947, + "kept_token_count": 571.4609375, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856771051883697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5948944282135923e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.4123229980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3966369628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10110, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.30032669603824613, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.12796785458922386, + "grad_norm": 0.0002718881005421281, + "kept_token_count": 572.5953125, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865294218063354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.26196450664429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.85300903320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9221527099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10120, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29948668777942655, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.11872760131955147, + "grad_norm": 0.00028164207469671965, + "kept_token_count": 573.0234375, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870269954204559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.067599231551867e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.0430847167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.53855285644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10130, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.3010076195001602, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.12722059860825538, + "grad_norm": 0.00023600917484145612, + "kept_token_count": 572.2921875, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862217247486115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.382164563343395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.217431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.04967041015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10140, + "valid_token_count": 818.7421875 + }, + { + "actual_token_mask_ratio": 0.2998518615961075, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.11806184351444245, + "grad_norm": 0.00023375748423859477, + "kept_token_count": 572.94375, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864793419837952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.28152577317087e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.699560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.62326965332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10150, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.299203622341156, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.1227114275097847, + "grad_norm": 0.0003976592852268368, + "kept_token_count": 573.39375, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859823048114776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.475681828102097e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.1337554931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.22797241210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10160, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.30069929361343384, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.13461276590824128, + "grad_norm": 0.0003126916999462992, + "kept_token_count": 572.3234375, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868082225322723, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.153058991709258e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.1793487548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3870391845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10170, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2997970014810562, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.13080291524529458, + "grad_norm": 0.00034266538568772376, + "kept_token_count": 572.94375, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856449902057648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6074457461363636e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.8642150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.92845764160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10180, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29952219128608704, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.12267126366496087, + "grad_norm": 0.0002823597169481218, + "kept_token_count": 573.1640625, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868364632129669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.14202441991074e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.21620483398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3098205566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10190, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.30020148754119874, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.1256919763982296, + "grad_norm": 0.00024502663291059434, + "kept_token_count": 572.8296875, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856601476669311, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.601527518592775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.0121765136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3124725341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10200, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30068619549274445, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.12911772429943086, + "grad_norm": 0.00021311531600076705, + "kept_token_count": 572.2953125, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866099953651428, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2304887503851205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.6743621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1884429931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10210, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.3001632481813431, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.12660786658525466, + "grad_norm": 0.00027098675491288304, + "kept_token_count": 572.6328125, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869776546955109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.086873134132475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.48467407226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9436370849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10220, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30031555593013765, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.11449693143367767, + "grad_norm": 0.00027967107598669827, + "kept_token_count": 572.3390625, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871234178543091, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.029927669966128e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.47449645996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.84351501464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10230, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.2993119776248932, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.12332321777939796, + "grad_norm": 0.00024917558766901493, + "kept_token_count": 573.4453125, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862140536308288, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.385154545365367e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.83289489746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7432556152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30012090802192687, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.12189097478985786, + "grad_norm": 0.0002872077457141131, + "kept_token_count": 572.609375, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867255926132202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1853346303687434e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.88112182617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.20538330078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10250, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3004906982183456, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.13222578465938567, + "grad_norm": 0.00027015269733965397, + "kept_token_count": 572.3484375, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866905748844147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1990085921715945e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.9210632324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.61174926757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10260, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.3002476006746292, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.14305634573101997, + "grad_norm": 0.00021530459343921393, + "kept_token_count": 572.5734375, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872162818908692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.993661132175475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.5036163330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1693450927734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10270, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2999837636947632, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.12676382511854173, + "grad_norm": 0.0003643448872026056, + "kept_token_count": 572.5578125, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855206251144409, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6560294979135506e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6291442871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3857788085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10280, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.30018418431282046, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.12515417411923407, + "grad_norm": 0.0002957165997941047, + "kept_token_count": 572.5234375, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867321908473968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.182761087780818e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.9897399902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.27586975097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10290, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2997164517641068, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.13053589165210724, + "grad_norm": 0.0003101626643911004, + "kept_token_count": 573.065625, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867395401000977, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1798885033349505e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.67210388183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3060668945312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10300, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2991323500871658, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.12861878499388696, + "grad_norm": 0.00036889041075482965, + "kept_token_count": 573.3796875, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9853274703025818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.731478631787468e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.70498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4543090820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10310, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30041553676128385, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.12663043588399886, + "grad_norm": 0.0002909461036324501, + "kept_token_count": 572.3234375, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855741620063782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.635112247546203e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.52585754394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4470275878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10320, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.29931601583957673, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.1263876773416996, + "grad_norm": 0.00026420014910399914, + "kept_token_count": 573.4515625, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859395623207092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.492382842930965e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.13031005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4461242675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10330, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.300720277428627, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.12571196183562278, + "grad_norm": 0.00038779847091063857, + "kept_token_count": 571.903125, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856009781360626, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.624640834867023e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.98476257324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6197540283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10340, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.29893364012241364, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.13171835020184516, + "grad_norm": 0.00028287965687923133, + "kept_token_count": 573.6046875, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858367502689361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.532534705707803e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.5208404541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5385467529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10350, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.3011521577835083, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.13121474906802177, + "grad_norm": 0.00022491905838251114, + "kept_token_count": 572.0109375, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865022957324981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.272562557365745e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.8588806152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.51316528320314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10360, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.3001027494668961, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.12237650454044342, + "grad_norm": 0.00023485660494770855, + "kept_token_count": 572.9234375, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867256164550782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.185328045627102e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.5326293945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1173919677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10370, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.30086526572704314, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.12725956961512566, + "grad_norm": 0.000316709018079564, + "kept_token_count": 572.1, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9855173230171204, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.657316869474016e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.13873291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.91571044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10380, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2998079568147659, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.13175444602966307, + "grad_norm": 0.00025482653290964663, + "kept_token_count": 572.946875, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868569374084473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.13402970682364e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.1697235107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70298461914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10390, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29931199848651885, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.12938774824142457, + "grad_norm": 0.00024195571313612163, + "kept_token_count": 573.653125, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870533227920533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0573171392898075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.49208374023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89219665527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10400, + "valid_token_count": 818.6984375 + }, + { + "actual_token_mask_ratio": 0.30056255459785464, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.12628141418099403, + "grad_norm": 0.00024759932421147823, + "kept_token_count": 572.3390625, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862491846084595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3714260502601974e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9391296386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6882843017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10410, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.29992947578430174, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.11947867572307587, + "grad_norm": 0.0002902860869653523, + "kept_token_count": 572.68125, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852351129055024, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.7675546850077807e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.47478942871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.15423278808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10420, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.30060438215732577, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.12537594586610795, + "grad_norm": 0.00034726940793916583, + "kept_token_count": 572.31875, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866039454936981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.232854055066127e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.9539398193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.97276306152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10430, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2991888463497162, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.12701417803764342, + "grad_norm": 0.000302758562611416, + "kept_token_count": 573.484375, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865444660186767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.256086951703765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.9359649658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.87745056152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10440, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30044980347156525, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.1326352320611477, + "grad_norm": 0.00030519257416017354, + "kept_token_count": 572.3171875, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863464593887329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3334360200096856e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.6646789550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9529113769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10450, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.3002578020095825, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.12675342112779617, + "grad_norm": 0.0003234940522816032, + "kept_token_count": 572.4109375, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875044107437134, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.881104687228799e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.2436218261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5159118652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10460, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.29977686405181886, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.11892049759626389, + "grad_norm": 0.00027069958741776645, + "kept_token_count": 572.978125, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867307901382446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.183301436773036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9914306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.99949645996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10470, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.29914850890636446, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.12606013119220733, + "grad_norm": 0.00027123745530843735, + "kept_token_count": 573.8234375, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867813110351562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.163572750461753e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.5535003662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5171173095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10480, + "valid_token_count": 818.753125 + }, + { + "actual_token_mask_ratio": 0.2997777134180069, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.13305791094899178, + "grad_norm": 0.0003375444211997092, + "kept_token_count": 572.9234375, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860174059867859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4619667935185134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.77903747558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.21045837402346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3011127322912216, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.14507163763046266, + "grad_norm": 0.0002819474320858717, + "kept_token_count": 571.8890625, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862698256969452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.363368909456767e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.9692810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.19475708007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10500, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30136646926403043, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.12641502171754837, + "grad_norm": 0.0003249119035899639, + "kept_token_count": 571.4953125, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856806397438049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.593519927060697e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.81792907714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2622802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10510, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2993976801633835, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.13348709046840668, + "grad_norm": 0.0002441184624331072, + "kept_token_count": 573.3796875, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986921751499176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.108712284709327e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.1934478759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8166809082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10520, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29959853291511535, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.13456401750445365, + "grad_norm": 0.0002923597930930555, + "kept_token_count": 573.1421875, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861428380012512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4129670024849476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.8471282958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3535675048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10530, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3006641805171967, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.13263383135199547, + "grad_norm": 0.00021659380581695586, + "kept_token_count": 572.11875, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863685309886933, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.324815138010308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.40972595214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7737213134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10540, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3005297988653183, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.13270285800099374, + "grad_norm": 0.000288203387754038, + "kept_token_count": 572.1078125, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9852759599685669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.751603130192962e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.5983917236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.26061096191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10550, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.29972176551818847, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.12910205125808716, + "grad_norm": 0.00023170167696662247, + "kept_token_count": 572.8890625, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872369289398193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.985592604498379e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.11937255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0338104248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10560, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2996662199497223, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.12978790029883386, + "grad_norm": 0.00030422056443057954, + "kept_token_count": 573.23125, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9859121024608613, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.503103311639279e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.60576782226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10570, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.3003639131784439, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.12133207768201829, + "grad_norm": 0.00027387498994357884, + "kept_token_count": 572.2515625, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867842793464661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.162410670891404e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.30120544433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10580, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.3000371605157852, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.13321698904037477, + "grad_norm": 0.0004025848757009953, + "kept_token_count": 572.7640625, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858263731002808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.5365918524330485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.1299285888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.97327880859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10590, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2988441675901413, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.1225006602704525, + "grad_norm": 0.0002973465307150036, + "kept_token_count": 573.921875, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864863753318787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.278777753119357e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.40321960449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7016296386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10600, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3008874148130417, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.12397009134292603, + "grad_norm": 0.000214026978937909, + "kept_token_count": 572.3234375, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874414384365082, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9057093929150144e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.94945373535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5618194580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10610, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.3018084794282913, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.14345332905650138, + "grad_norm": 0.0002483994758222252, + "kept_token_count": 571.375, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866059005260468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.23209626408061e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.31207885742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3436248779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10620, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3008242130279541, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.1400538206100464, + "grad_norm": 0.0002912890922743827, + "kept_token_count": 571.95625, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867394685745239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1799139691866e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.292138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9245300292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10630, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.3005644142627716, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.13505831435322763, + "grad_norm": 0.0002251267433166504, + "kept_token_count": 572.371875, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863578081130981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.328997649485245e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.3869995117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7862823486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10640, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3007275342941284, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.13942612931132317, + "grad_norm": 0.00022361053561326116, + "kept_token_count": 572.25625, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867932558059692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.158902458788361e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.4018218994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.88734436035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10650, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2997484773397446, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.11847557425498963, + "grad_norm": 0.00038287288043648005, + "kept_token_count": 572.9125, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865070760250092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.27069245435996e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.375732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3575958251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10660, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.299527582526207, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.13110895678400994, + "grad_norm": 0.00023262994363904, + "kept_token_count": 573.0078125, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868931770324707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1198738947277886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.2072021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6123474121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10670, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.29971790611743926, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.14132973849773406, + "grad_norm": 0.0002865730202756822, + "kept_token_count": 572.9359375, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863034188747406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.350249593902845e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.6192932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.23907165527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10680, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30007500350475313, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.12351250648498535, + "grad_norm": 0.00033509513013996184, + "kept_token_count": 573.053125, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856275737285614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.6142407265724616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.52012939453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5874847412109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10690, + "valid_token_count": 818.728125 + }, + { + "actual_token_mask_ratio": 0.30075508952140806, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.13211403265595437, + "grad_norm": 0.00023038560175336897, + "kept_token_count": 572.275, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.98672896027565, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.184017754800152e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.6521331787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8105010986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10700, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2992822229862213, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.13073616325855256, + "grad_norm": 0.00024811478215269744, + "kept_token_count": 573.2875, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864620983600616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.288264437695034e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.9578826904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.62294616699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10710, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29957776963710786, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.13210608139634133, + "grad_norm": 0.0002537975669838488, + "kept_token_count": 572.8875, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866826295852661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.202114443818573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.19032897949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.26466674804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10720, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.2992165327072144, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.1323749080300331, + "grad_norm": 0.00037810145295225084, + "kept_token_count": 573.2828125, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861984550952911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3912445218884385e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.4736053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8317138671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10730, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29957116544246676, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.13185106441378594, + "grad_norm": 0.00022600439842790365, + "kept_token_count": 573.0828125, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986876380443573, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.126430842210539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.63584594726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.21019287109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10740, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.300799560546875, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.13278234377503395, + "grad_norm": 0.00024726291303522885, + "kept_token_count": 572.16875, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986913126707077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1120814532623626e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.34556884765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.83116149902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10750, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.29942659437656405, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.13088109940290452, + "grad_norm": 0.0002909078903030604, + "kept_token_count": 573.1515625, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986373782157898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3227608441375195e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.1681427001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.85303955078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10760, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.3001495569944382, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.11993656605482102, + "grad_norm": 0.00023255933774635196, + "kept_token_count": 572.4453125, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866759657859803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.204728040553164e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.7836883544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.48732299804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10770, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.2997363775968552, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.1199134849011898, + "grad_norm": 0.00018920039292424917, + "kept_token_count": 573.146875, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868521809577941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1358825658098795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.03013305664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.16001892089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2988336026668549, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.12221846878528594, + "grad_norm": 0.0002603501779958606, + "kept_token_count": 573.69375, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987442284822464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.905374116788152e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.11963806152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0174896240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10790, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3002682149410248, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.1433626376092434, + "grad_norm": 0.00032377790194004774, + "kept_token_count": 572.6203125, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.985302186012268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.741349159507081e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.7698089599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9128143310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2991846352815628, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.13227961510419844, + "grad_norm": 0.00027394932112656534, + "kept_token_count": 573.5640625, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870991408824921, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0394165373290886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.40447692871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8367034912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10810, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.29896496832370756, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.12743212431669235, + "grad_norm": 0.00022416847059503198, + "kept_token_count": 573.5984375, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866896450519562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.199377119424753e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.4340484619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9228454589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10820, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3004991918802261, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.1342218779027462, + "grad_norm": 0.00026086560683324933, + "kept_token_count": 572.6078125, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871160566806794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.03280978591647e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.7042175292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.60520935058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10830, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.3000850766897202, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.13117487132549285, + "grad_norm": 0.00023856064944993705, + "kept_token_count": 572.7921875, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860528588294983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.448121955851093e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.7436584472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6750793457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10840, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30075581967830656, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.13176528811454774, + "grad_norm": 0.0002732306602410972, + "kept_token_count": 572.2875, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871960401535034, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.001565332349855e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.40675354003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.28204650878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10850, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30075842440128325, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.12859400510787963, + "grad_norm": 0.0004470946441870183, + "kept_token_count": 572.2796875, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867350161075592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.18164895765949e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.4317626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.58595275878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10860, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.29960196912288667, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.12163891345262527, + "grad_norm": 0.0002727716928347945, + "kept_token_count": 573.1828125, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866614520549775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2103887355769984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.23448181152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0375915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10870, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3003430932760239, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.1348831109702587, + "grad_norm": 0.0002978577686008066, + "kept_token_count": 572.7015625, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870385944843292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0630631812964565e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.925927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6574737548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10880, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.30098343193531035, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.12254942432045937, + "grad_norm": 0.00025674852076917887, + "kept_token_count": 572.0671875, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861885368824005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3951197332935405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.9737609863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.92962951660155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2990573614835739, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.11819066181778907, + "grad_norm": 0.0003641000366769731, + "kept_token_count": 573.5359375, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861822664737702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.397568929765839e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.14146728515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4770782470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10900, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3011091351509094, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.13001034632325173, + "grad_norm": 0.0002480717666912824, + "kept_token_count": 571.7765625, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866848528385163, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.201248459343333e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.72950744628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9615905761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2995434731245041, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.12744685038924217, + "grad_norm": 0.00027424839208833873, + "kept_token_count": 573.3390625, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870852768421173, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.04483345139306e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.1943664550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.22312622070314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10920, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.2994100123643875, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.1314658872783184, + "grad_norm": 0.0003509504022076726, + "kept_token_count": 573.2890625, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856659412384033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.599256546702236e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.3708770751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4811309814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10930, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3004910498857498, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.13573777824640273, + "grad_norm": 0.00021619077597279102, + "kept_token_count": 572.6421875, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876155972480773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.837676569877658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.5442718505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5732391357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10940, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.29979813992977145, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.1315576672554016, + "grad_norm": 0.00031675613718107343, + "kept_token_count": 573.0484375, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9857072532176971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.58312596695032e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.7526489257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.41343994140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10950, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30003940761089326, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.12910312488675119, + "grad_norm": 0.00023258087458088994, + "kept_token_count": 572.9625, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850230932235717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.850369889230933e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.67042541503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7412078857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10960, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.30055764615535735, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.13700497895479202, + "grad_norm": 0.00022178080689627677, + "kept_token_count": 572.3234375, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871411144733429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0230130727868524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.7368133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.91229248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10970, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.3001537948846817, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.11475526168942451, + "grad_norm": 0.00032607594039291143, + "kept_token_count": 572.678125, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868180155754089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.149229109520092e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.79295959472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0453674316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2992922216653824, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.13241981938481331, + "grad_norm": 0.00033463272848166525, + "kept_token_count": 573.4046875, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860235929489136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.4595537221757695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.8688995361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8528228759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3003630697727203, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.12589670568704606, + "grad_norm": 0.0003466096241027117, + "kept_token_count": 572.5390625, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868557989597321, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.13447419507429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.0987152099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89695129394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11000, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2996034026145935, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.12987195327877998, + "grad_norm": 0.0003082635230384767, + "kept_token_count": 573.0625, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868335962295532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1431470274110326e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.5979827880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11010, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29985622465610506, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.13951915353536606, + "grad_norm": 0.00022250390611588955, + "kept_token_count": 573.0484375, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869304955005646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.105287091282662e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.90855712890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9308258056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11020, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.30035279095172884, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.12294597551226616, + "grad_norm": 0.00022056870511732996, + "kept_token_count": 572.5046875, + "learning_rate": 2.245843529059e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9861894309520721, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.394772888394073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.12583312988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.235009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11030, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2986262381076813, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.12991802021861076, + "grad_norm": 0.0002799200592562556, + "kept_token_count": 573.9203125, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860264003276825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.458455307234544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9542602539062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.74693603515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11040, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.2995899736881256, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.13898975625634194, + "grad_norm": 0.00022182361863087863, + "kept_token_count": 573.1171875, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986217337846756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.383871648518834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.68817443847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.90818176269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11050, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29926398396492004, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.12534528747200965, + "grad_norm": 0.00027753549511544406, + "kept_token_count": 573.59375, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864681780338287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.285889637889341e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.5825500488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75205383300784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11060, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.2997943788766861, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.1280694954097271, + "grad_norm": 0.00026366571546532214, + "kept_token_count": 572.8921875, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866062641143799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2319430324132554e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.70481872558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.49033813476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11070, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.299762436747551, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.12944951504468918, + "grad_norm": 0.0002636006101965904, + "kept_token_count": 572.81875, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874606609344483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8981946383719334e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.4791625976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3144897460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11080, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.30083209574222564, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.13208430036902427, + "grad_norm": 0.0002776674518827349, + "kept_token_count": 572.0296875, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869391739368438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.101905262563378e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.2892639160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9623565673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11090, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.2997880905866623, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.12979219481348991, + "grad_norm": 0.00021576181461568922, + "kept_token_count": 572.7359375, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987317955493927, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.953943753207568e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.00001831054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6518951416016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11100, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.30040628612041476, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.12158445790410041, + "grad_norm": 0.0002112002403009683, + "kept_token_count": 572.378125, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869539320468903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.096137210784946e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.0758758544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2231842041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11110, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2998047798871994, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.13451194912195205, + "grad_norm": 0.00022420004825107753, + "kept_token_count": 572.91875, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873467206954956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9426988334744236e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3655242919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2063751220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11120, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30060812532901765, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.1466626100242138, + "grad_norm": 0.00021702087542507797, + "kept_token_count": 572.0640625, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860007047653199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.46849663805915e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.43646240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0690124511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11130, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.29872254431247713, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.12829158157110215, + "grad_norm": 0.00031950187985785306, + "kept_token_count": 573.7109375, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875307679176331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.870809534622822e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.87066650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0740020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11140, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.2995900630950928, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.12171076610684395, + "grad_norm": 0.00023786007659509778, + "kept_token_count": 573.125, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987632018327713, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.831260521314107e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.4805938720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.25360107421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11150, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2997326791286469, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.1349988639354706, + "grad_norm": 0.0002824918774422258, + "kept_token_count": 572.9921875, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872134327888489, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.994771006749943e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.0493682861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0832550048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11160, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.3003320187330246, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.13283531069755555, + "grad_norm": 0.0002212446997873485, + "kept_token_count": 572.26875, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872765243053436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9701299576554445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9238311767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.55816040039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11170, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.2999467015266418, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.12820446789264678, + "grad_norm": 0.0003250224981456995, + "kept_token_count": 572.796875, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868353366851806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1424639241304246e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.8815948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2672882080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11180, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.30052330493927004, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.1419661708176136, + "grad_norm": 0.00030458703986369073, + "kept_token_count": 572.359375, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986521327495575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.265118452371098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2458190917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6434661865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11190, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3004873663187027, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.12693366259336472, + "grad_norm": 0.00038305908674374223, + "kept_token_count": 572.3421875, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866330087184906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.221497813181486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.6426239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2783996582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3008202314376831, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.1309698276221752, + "grad_norm": 0.00027247972320765257, + "kept_token_count": 571.95625, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868717730045319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1282319691381415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.4493408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2489349365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11210, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.3009665638208389, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.11701067760586739, + "grad_norm": 0.0002412255125818774, + "kept_token_count": 572.01875, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987500661611557, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8825701014720835e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.08095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.75433654785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11220, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30024928152561187, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.13137743696570398, + "grad_norm": 0.00036573418765328825, + "kept_token_count": 572.3046875, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986400431394577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3123532416066156e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.3232360839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.58002319335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11230, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.29993794560432435, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.13246433287858964, + "grad_norm": 0.0002267950912937522, + "kept_token_count": 572.6671875, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9850442588329316, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.842113168910146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.11700744628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.00867614746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11240, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2997170478105545, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.13663508892059326, + "grad_norm": 0.00022115280444268137, + "kept_token_count": 572.74375, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860000789165497, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.468735980684869e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.13782653808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7776672363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11250, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.299897289276123, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.12338838651776314, + "grad_norm": 0.00027838745154440403, + "kept_token_count": 572.946875, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863134980201721, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.346309881133493e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.6831298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1903350830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11260, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.29958897531032563, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.12702207639813423, + "grad_norm": 0.00034913583658635616, + "kept_token_count": 573.05625, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868455231189728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1384856124059294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.0059051513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0690643310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11270, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.30124941766262053, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.1221965380012989, + "grad_norm": 0.0003566524828784168, + "kept_token_count": 572.0140625, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870793998241425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.047129816375673e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.5875946044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.84576721191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11280, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.30032919347286224, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.12257329598069192, + "grad_norm": 0.00030114492983557284, + "kept_token_count": 572.571875, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987733119726181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7917734991642644e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.8654479980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3289031982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11290, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30091876685619356, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.12003220692276954, + "grad_norm": 0.00023448340652976185, + "kept_token_count": 572.140625, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869775652885437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0869101687567306e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.50159912109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5929321289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11300, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2995161205530167, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.12479297891259193, + "grad_norm": 0.00027537986170500517, + "kept_token_count": 573.4265625, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866950333118438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.19727520440938e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.64743957519534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.39324951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11310, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.2993087887763977, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.12455143555998802, + "grad_norm": 0.00022681671543978155, + "kept_token_count": 573.4984375, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867618381977081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.171175689611118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.08134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2436492919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11320, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2992437094449997, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.13428740128874778, + "grad_norm": 0.00018767283472698182, + "kept_token_count": 573.840625, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987632554769516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8310536294593474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.30857543945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1079956054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11330, + "valid_token_count": 818.890625 + }, + { + "actual_token_mask_ratio": 0.30090085566043856, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.13537421599030494, + "grad_norm": 0.000292118958896026, + "kept_token_count": 571.9125, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9856249809265136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.61526063393103e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.8434204101562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6745666503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11340, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.3001490026712418, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.13516876995563507, + "grad_norm": 0.00026400742353871465, + "kept_token_count": 572.61875, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869693040847778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0901356735266746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.3289733886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6239410400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11350, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.3006612867116928, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.12711210697889327, + "grad_norm": 0.0003614829038269818, + "kept_token_count": 572.2078125, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9851293087005615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.8088814694201576e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.41661682128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6960205078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11360, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29964154958724976, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.12812438011169433, + "grad_norm": 0.0002596070698928088, + "kept_token_count": 573.44375, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872178614139557, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9930431850953026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.4147888183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.41705627441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11370, + "valid_token_count": 818.778125 + }, + { + "actual_token_mask_ratio": 0.3008234053850174, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.13120242431759835, + "grad_norm": 0.0002702329366002232, + "kept_token_count": 572.2375, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869139969348908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.11173864651937e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.32265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.23353271484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11380, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.29966045916080475, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.1304802492260933, + "grad_norm": 0.00026890827575698495, + "kept_token_count": 573.1953125, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865760266780853, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.243764026090503e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.4148773193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1634033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2999427318572998, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.13277106881141662, + "grad_norm": 0.00024411649792455137, + "kept_token_count": 572.90625, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874640882015229, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.896858554275241e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.36033325195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9551971435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11400, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.2994465887546539, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.12672922611236573, + "grad_norm": 0.00025603201356716454, + "kept_token_count": 573.4703125, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9862420678138732, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.374210013542324e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1292388916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11410, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.2984600573778152, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.12149159759283065, + "grad_norm": 0.0003322176053188741, + "kept_token_count": 574.0125, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871585249900818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.016215291107073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.4037658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.35834045410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11420, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29966908395290376, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.12521782591938974, + "grad_norm": 0.00021509146608877927, + "kept_token_count": 572.8421875, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880224525928497, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.678748773585539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.28200378417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.95252380371096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11430, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.299114727973938, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.11704735308885575, + "grad_norm": 0.0003502508334349841, + "kept_token_count": 573.684375, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876609325408936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.819965397473425e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.7361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.48822021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11440, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.3009774208068848, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.1303958036005497, + "grad_norm": 0.00021721620578318834, + "kept_token_count": 571.834375, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873053193092346, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.958878125762567e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.68150634765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4688262939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11450, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.3005865514278412, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.12606664523482322, + "grad_norm": 0.0002795469481498003, + "kept_token_count": 572.321875, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863363027572631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3373969785752705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.40240783691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1632446289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2997101128101349, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.12645241245627403, + "grad_norm": 0.0003365768352523446, + "kept_token_count": 573.146875, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872715830802917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.97205663123168e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2070709228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61004333496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11470, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2992739140987396, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.12831753492355347, + "grad_norm": 0.00021895371901337057, + "kept_token_count": 573.1125, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871589958667755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0160286991740574e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.8431396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.86187133789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11480, + "valid_token_count": 817.8890625 + }, + { + "actual_token_mask_ratio": 0.2994562864303589, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.13034872561693192, + "grad_norm": 0.00025762440054677427, + "kept_token_count": 573.3125, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873161971569061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.954629148414824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1924713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.83929138183595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11490, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3007724732160568, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.12726602032780648, + "grad_norm": 0.000254124344792217, + "kept_token_count": 572.1203125, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868783414363861, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1256669758004135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.0370361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.07843627929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11500, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.299149489402771, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.12157186567783355, + "grad_norm": 0.00030557974241673946, + "kept_token_count": 573.4828125, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874244928359985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.912326076009776e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.05475463867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6864349365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11510, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30007335245609285, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.13471893295645715, + "grad_norm": 0.0003586207749322057, + "kept_token_count": 572.8390625, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864071547985077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.309722364472691e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.63869018554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.02422485351565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11520, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3000495105981827, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.13603687807917594, + "grad_norm": 0.00022096632164902985, + "kept_token_count": 572.7640625, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873964488506317, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.92327926622238e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.9696868896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44911499023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11530, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30018678605556487, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.13145517259836198, + "grad_norm": 0.0005427161231637001, + "kept_token_count": 572.7359375, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868137300014496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1509036711649967e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4944610595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5968048095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11540, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.3005894869565964, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.13734753653407097, + "grad_norm": 0.0002381453668931499, + "kept_token_count": 572.346875, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869964182376861, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.079543770989403e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.20575561523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.39491882324216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11550, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29975748956203463, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.12598087564110755, + "grad_norm": 0.0002459135139361024, + "kept_token_count": 573.2296875, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875571846961975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8604955736664124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9473876953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1604248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11560, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.3012185662984848, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.12714640870690347, + "grad_norm": 0.0003667159762699157, + "kept_token_count": 571.8203125, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872601449489593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.976525488018524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.25645446777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1605285644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11570, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.30079394578933716, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.12910228669643403, + "grad_norm": 0.00019337692356202751, + "kept_token_count": 572.065625, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876571416854858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8214470371021886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4074340820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.84253845214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11580, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2994642734527588, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.13572858572006224, + "grad_norm": 0.00023231105296872556, + "kept_token_count": 573.078125, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878502011299133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7460291534662245e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.73448486328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.12159729003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11590, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.30072527527809145, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.12421688660979271, + "grad_norm": 0.00027255716850049794, + "kept_token_count": 572.3484375, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869178891181946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1102177894790654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.9949066162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.81850280761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11600, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.3004115432500839, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.14129041731357575, + "grad_norm": 0.00024898650008253753, + "kept_token_count": 572.70625, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865048170089722, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.271575282677077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.80726318359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.49866943359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11610, + "valid_token_count": 818.63125 + }, + { + "actual_token_mask_ratio": 0.30041222274303436, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.13563481271266936, + "grad_norm": 0.00024195208970922977, + "kept_token_count": 572.496875, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860675930976868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.442363253678195e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.3542236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.40862426757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11620, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29906055629253386, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.13809703812003135, + "grad_norm": 0.00027930823853239417, + "kept_token_count": 573.5765625, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865912914276123, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2377953034010714e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.5351043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.31573181152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3002031683921814, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.13292451426386834, + "grad_norm": 0.00037129613338038325, + "kept_token_count": 572.5296875, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875749766826629, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.853542595810723e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.24208679199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.480126953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11640, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.30038333535194395, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.12861964404582976, + "grad_norm": 0.00023806793615221977, + "kept_token_count": 572.6171875, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987935596704483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.712676964118146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.52479248046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5404815673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11650, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.30027173161506654, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.12895229011774062, + "grad_norm": 0.00022463452478405088, + "kept_token_count": 572.6, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874032497406006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9206180483452046e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5936798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.73701171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11660, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.3008750557899475, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.1345822848379612, + "grad_norm": 0.00020879933435935527, + "kept_token_count": 572.021875, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871640384197236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.014061571273487e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.7020568847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7021789550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11670, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3009554177522659, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.12450773864984513, + "grad_norm": 0.00026761129265651107, + "kept_token_count": 571.990625, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870197355747223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.070434963272419e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2809692382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.71507263183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11680, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30060969293117523, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.13574683889746667, + "grad_norm": 0.00025503471260890365, + "kept_token_count": 572.3390625, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875068843364716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8801432421896605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7371826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.03839111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3008995682001114, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.12225591614842415, + "grad_norm": 0.00028319263947196305, + "kept_token_count": 572.0015625, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869977593421936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.079017537354957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.6231384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4119903564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11700, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2995529264211655, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.13433128148317336, + "grad_norm": 0.0002478896640241146, + "kept_token_count": 572.9578125, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874782681465148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.891323151241522e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.5082092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2785339355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11710, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.3002044677734375, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.1273833192884922, + "grad_norm": 0.00023336180311162025, + "kept_token_count": 572.540625, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871244072914124, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.029547501180787e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7749877929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.28769836425784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11720, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2999488592147827, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.12886817306280135, + "grad_norm": 0.0002584012399893254, + "kept_token_count": 572.940625, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986810827255249, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.152041521796491e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.71021728515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.19378356933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11730, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.2995471298694611, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.1247191920876503, + "grad_norm": 0.00024704192765057087, + "kept_token_count": 573.215625, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871704697608947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.011553585063666e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.80617370605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.92140197753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11740, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2980834871530533, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.13839124888181686, + "grad_norm": 0.00029406193061731756, + "kept_token_count": 574.3125, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869082689285278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1139829884050415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.65277404785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.62693481445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11750, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29921028912067416, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.12537871822714805, + "grad_norm": 0.0003002846788149327, + "kept_token_count": 573.5359375, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866265714168548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.224017149885185e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.1906341552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.29872436523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11760, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3003643810749054, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.1337779462337494, + "grad_norm": 0.0001932506711455062, + "kept_token_count": 572.4578125, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874261915683746, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.911664618703071e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.87572021484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7104064941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11770, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3012007296085358, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.12202258035540581, + "grad_norm": 0.0002040555264102295, + "kept_token_count": 571.9796875, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878380477428437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7507828639936633e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.5909088134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.18590087890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.3005809515714645, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.13084446862339974, + "grad_norm": 0.0003049387887585908, + "kept_token_count": 572.3203125, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867093980312347, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1916544543928465e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.6816833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.69776306152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11790, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2998466670513153, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.13074245750904084, + "grad_norm": 0.00028349089552648365, + "kept_token_count": 573.04375, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986929738521576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.105589771119412e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7520263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6648376464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11800, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.3000897794961929, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.134230837225914, + "grad_norm": 0.00026956506189890206, + "kept_token_count": 572.7390625, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873805105686188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.92949959152611e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.9870880126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4549896240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11810, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2996060401201248, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.14225212186574937, + "grad_norm": 0.00033314689062535763, + "kept_token_count": 573.03125, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865631461143494, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2487931679934266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.63536682128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5467468261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11820, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.29941887259483335, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.12781226709485055, + "grad_norm": 0.0002801944501698017, + "kept_token_count": 573.3703125, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9860803842544555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.437368527054787e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.5707214355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8946838378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11830, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29917181134223936, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.13410717472434044, + "grad_norm": 0.00035484740510582924, + "kept_token_count": 573.30625, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866841912269593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.2015033361385576e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.32098083496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4491668701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11840, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.3009327590465546, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.13398877903819084, + "grad_norm": 0.00034203630639240146, + "kept_token_count": 572.0375, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986185896396637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.396150991145987e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7165222167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.32086791992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11850, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3003924250602722, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.1317538134753704, + "grad_norm": 0.0003060765448026359, + "kept_token_count": 572.509375, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871351242065429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0253620793228036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.79355773925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.15391540527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11860, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.29993127584457396, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.12441059425473214, + "grad_norm": 0.00025152447051368654, + "kept_token_count": 573.0625, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873454988002777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.94318603159627e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.87896728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.44358520507814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11870, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.3002306342124939, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.13532458171248435, + "grad_norm": 0.00025193762849085033, + "kept_token_count": 572.571875, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874656975269318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8962327855406326e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.83506774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01304321289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11880, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.299675264954567, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.13554508984088898, + "grad_norm": 0.00021386941079981625, + "kept_token_count": 573.14375, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869074165821076, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1143099699402225e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.0932983398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2469146728516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11890, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.30001933574676515, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.13069833293557168, + "grad_norm": 0.00041114765917882323, + "kept_token_count": 572.875, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871863961219788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0053370068781077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1317199707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.89267578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11900, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29922441244125364, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.12220694869756699, + "grad_norm": 0.0002812264137901366, + "kept_token_count": 573.3921875, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865997970104218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.23447830346413e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.3480163574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.02025146484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11910, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.2998718559741974, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.13782552182674407, + "grad_norm": 0.00028893721173517406, + "kept_token_count": 572.9515625, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868892550468444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.12140104547143e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.33481750488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43712463378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11920, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30068025290966033, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.138946745544672, + "grad_norm": 0.00021736774942837656, + "kept_token_count": 572.0203125, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987664008140564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.818770648853388e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.7200531005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.83486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11930, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.30065331161022185, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.13604059293866158, + "grad_norm": 0.0002511226339265704, + "kept_token_count": 572.25625, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868999481201172, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.11722424562322e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.5074798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.46714172363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11940, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2994573175907135, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.13162808120250702, + "grad_norm": 0.0003096795699093491, + "kept_token_count": 573.3171875, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873638987541199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.935996767017059e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8167388916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.97136840820315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.3004711538553238, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.13307542577385903, + "grad_norm": 0.000283819274045527, + "kept_token_count": 572.3078125, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868214964866638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.147871452209074e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.41994018554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61205139160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11960, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.3005427211523056, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.12661759331822395, + "grad_norm": 0.00028773912345059216, + "kept_token_count": 572.4703125, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868062973022461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1538111438276243e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.94708251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7336761474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11970, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.29898596107959746, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.13346003219485283, + "grad_norm": 0.0002580760628916323, + "kept_token_count": 573.8421875, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868427574634552, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.139567365404218e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.5275512695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01404724121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11980, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.3009311109781265, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.12073598429560661, + "grad_norm": 0.0002636364661157131, + "kept_token_count": 571.8984375, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873991727828979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.922219050058629e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.93964233398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1242156982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11990, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.30020399689674376, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.135108083486557, + "grad_norm": 0.00020989136828575283, + "kept_token_count": 572.7171875, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876484334468841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.824850584554952e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3534881591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.75322875976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12000, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.30041379630565646, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.14102624580264092, + "grad_norm": 0.00019359152065590024, + "kept_token_count": 572.3625, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870065331459046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.075591616332531e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4536865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7150939941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12010, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30028732419013976, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.12309738919138909, + "grad_norm": 0.00037416693521663547, + "kept_token_count": 572.5515625, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871957778930665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.001667668693699e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7018707275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72576599121095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12020, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3007478445768356, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.13091560825705528, + "grad_norm": 0.00024045641475822777, + "kept_token_count": 571.9265625, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868371069431305, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.141776637174189e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.24964599609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.49598693847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12030, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.3000254899263382, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.12728922590613365, + "grad_norm": 0.00020020740339532495, + "kept_token_count": 572.69375, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987826657295227, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7552315299981275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2073638916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.91040954589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3004596710205078, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.12757991254329681, + "grad_norm": 0.0002571964869275689, + "kept_token_count": 572.5015625, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876496195793152, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.824390598514583e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.0372009277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4679290771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3000577986240387, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.13593599274754525, + "grad_norm": 0.0003371944767422974, + "kept_token_count": 572.86875, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867384552955627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1803115275106394e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.64618225097655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4296569824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12060, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.30085821747779845, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.13074845522642137, + "grad_norm": 0.00030509609496220946, + "kept_token_count": 571.9578125, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873961687088013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.923394335492049e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2539459228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.53904113769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12070, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30032632648944857, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.12282838821411132, + "grad_norm": 0.0003761394473258406, + "kept_token_count": 572.6796875, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9863338589668273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.3383534759632314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9112945556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4365631103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12080, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.30052128434181213, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.13401429653167723, + "grad_norm": 0.00023409826098941267, + "kept_token_count": 572.4078125, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871110677719116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.034762143623084e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.6373626708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0713592529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12090, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3010392040014267, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.14112143218517303, + "grad_norm": 0.00021454767556861043, + "kept_token_count": 571.9578125, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867996156215668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.156421466381289e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8019287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1410339355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12100, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2996872991323471, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.13548185527324677, + "grad_norm": 0.0001644339645281434, + "kept_token_count": 572.80625, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874907851219177, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.88642806885764e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5095275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.90553894042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12110, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.3003008127212524, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.11763227209448815, + "grad_norm": 0.00029748378437943757, + "kept_token_count": 572.6953125, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873936772346497, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.924365748593118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8716094970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7480804443359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12120, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.3006391704082489, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.13359633684158326, + "grad_norm": 0.00026540798717178404, + "kept_token_count": 572.384375, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867696821689605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.168114003026858e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.2422790527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.86131591796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12130, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.30075140595436095, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.12016645818948746, + "grad_norm": 0.0003397107357159257, + "kept_token_count": 572.2125, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866389214992524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.219191953074187e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.39891662597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0193634033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12140, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2992796778678894, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.12447227835655213, + "grad_norm": 0.00023032614262774587, + "kept_token_count": 573.38125, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878138244152069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7602428458048963e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.0325134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.956787109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12150, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.30045305788516996, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.1323450103402138, + "grad_norm": 0.00027497654082253575, + "kept_token_count": 572.315625, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987428605556488, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9107201266451736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.0405731201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7088104248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12160, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2999298542737961, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.12842627391219139, + "grad_norm": 0.00043904277845285833, + "kept_token_count": 572.6671875, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867225110530853, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.186536109249573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8837036132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89149169921876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12170, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.3002226650714874, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.13939610943198205, + "grad_norm": 0.00032539665699005127, + "kept_token_count": 572.66875, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987673306465149, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8151308146771046e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2375183105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0841918945313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12180, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2998023271560669, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.12604696154594422, + "grad_norm": 0.00028632310568355024, + "kept_token_count": 573.03125, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865883529186249, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.238939338596538e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.47240295410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4887634277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12190, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.3009540557861328, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.1352010779082775, + "grad_norm": 0.00034449659870006144, + "kept_token_count": 571.95, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872370541095734, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9855489851324816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.77783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6689788818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12200, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29915620386600494, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.13637830913066865, + "grad_norm": 0.0002932535426225513, + "kept_token_count": 573.4265625, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874559581279755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.900037129118573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7521240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5520446777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12210, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29991055727005006, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.1289645217359066, + "grad_norm": 0.0002546699543017894, + "kept_token_count": 572.8703125, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987653237581253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.822973678528797e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.81093139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6093505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12220, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.2989057242870331, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.12787790596485138, + "grad_norm": 0.00021350923634599894, + "kept_token_count": 573.65625, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871264159679413, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0287615158595146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.10369262695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.92462463378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12230, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29957074820995333, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.12601394429802895, + "grad_norm": 0.0002751370193436742, + "kept_token_count": 573.046875, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873217523097992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.952461749780923e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.523779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.17632446289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12240, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.30052621066570284, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.12927723303437233, + "grad_norm": 0.000262951449258253, + "kept_token_count": 572.390625, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877942323684692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.767903446918353e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.67579650878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.302587890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30000847280025483, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.1233181819319725, + "grad_norm": 0.0003701084933709353, + "kept_token_count": 572.803125, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9858361005783081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.532798568310682e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.02705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1280090332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12260, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.29928180277347566, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.13529788032174112, + "grad_norm": 0.00021854476653970778, + "kept_token_count": 573.621875, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872599720954895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.976593299943488e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.21513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0310363769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12270, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.2995066612958908, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.12687084153294564, + "grad_norm": 0.00021609579562209547, + "kept_token_count": 573.4703125, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869590878486634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.094128100608941e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.15343322753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.25938720703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12280, + "valid_token_count": 818.6703125 + }, + { + "actual_token_mask_ratio": 0.29960009157657624, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.1256838783621788, + "grad_norm": 0.00027252838481217623, + "kept_token_count": 573.31875, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874829292297364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.889502852165606e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.27549133300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4594360351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12290, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.30036636590957644, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.135942891985178, + "grad_norm": 0.000250570650678128, + "kept_token_count": 572.68125, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880616068840027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.663453728426248e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4555694580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.66315307617185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12300, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.29941536784172057, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.13576074615120887, + "grad_norm": 0.0002515225496608764, + "kept_token_count": 573.259375, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868139743804931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.150809338374529e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.53798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.72662048339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12310, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2985993504524231, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.1314084380865097, + "grad_norm": 0.00030852577765472233, + "kept_token_count": 573.7390625, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869021832942962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1163526586606166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.21900634765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3019622802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12320, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.2997830450534821, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.12381556853652001, + "grad_norm": 0.00038333472912199795, + "kept_token_count": 573.225, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.987100237607956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.038987728767097e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.4194854736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8696228027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12330, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.3008868873119354, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.12304752096533775, + "grad_norm": 0.00022665801225230098, + "kept_token_count": 571.95625, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873323798179626, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.948309688188601e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.3887603759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8004974365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12340, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.29911766946315765, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.13542087748646736, + "grad_norm": 0.00025611845194362104, + "kept_token_count": 573.4671875, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876051664352417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8417524885735476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.632421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1063262939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29983564615249636, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.1321090765297413, + "grad_norm": 0.0002453204069752246, + "kept_token_count": 573.203125, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871494054794312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.019784985051956e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.72621765136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5653839111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12360, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.2998649626970291, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.12055130451917648, + "grad_norm": 0.00017800800560507923, + "kept_token_count": 572.959375, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9864494502544403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.293205431371462e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1357360839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5204345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12370, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30034259259700774, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.12573186606168746, + "grad_norm": 0.0002670006069820374, + "kept_token_count": 572.7140625, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875566601753235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.860700246354099e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2919616699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9828765869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12380, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.300354939699173, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.122310970723629, + "grad_norm": 0.00026880027144216, + "kept_token_count": 572.496875, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873220801353455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9523299094289544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9039031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5568542480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12390, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.2987213581800461, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.12734024599194527, + "grad_norm": 0.00028849602676928043, + "kept_token_count": 574.1296875, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878769814968109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7355717833852394e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.00487060546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5363037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12400, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.30004650056362153, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.12204181179404258, + "grad_norm": 0.00024994401610456407, + "kept_token_count": 573.065625, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876775979995728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8134548342204654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.80707397460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.97158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12410, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.3007068902254105, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.11862872764468194, + "grad_norm": 0.00025014206767082214, + "kept_token_count": 572.1140625, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880483269691467, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.668643923650961e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.7595916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1916900634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12420, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.3008119463920593, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.13519557043910027, + "grad_norm": 0.0002277788589708507, + "kept_token_count": 571.9921875, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871004104614258, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.038922390667722e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.13758850097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.33343200683595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12430, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.30064770579338074, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.11734602451324463, + "grad_norm": 0.0002628052025102079, + "kept_token_count": 572.515625, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876816511154175, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.811876169696916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0821899414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8484161376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12440, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.300648432970047, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.13534319251775742, + "grad_norm": 0.0003026540798600763, + "kept_token_count": 572.378125, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870427489280701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.061449483036995e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3102172851562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0675903320313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12450, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.30037873089313505, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.12723007425665855, + "grad_norm": 0.0002522182185202837, + "kept_token_count": 572.6171875, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884162008762359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5249440154293553e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3729278564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.24387512207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12460, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2998212069272995, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.13448075354099273, + "grad_norm": 0.0002820468507707119, + "kept_token_count": 573.04375, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878142356872559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.760083502333146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8502471923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2273376464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12470, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.3006756156682968, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.1271784208714962, + "grad_norm": 0.0002466800797265023, + "kept_token_count": 572.0953125, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879582762718201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7038198681548236e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7091979980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7999755859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12480, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3005236774682999, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.12695759162306786, + "grad_norm": 0.0002039941755356267, + "kept_token_count": 572.3984375, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880236268043519, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.678287441493012e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.8940765380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.05377197265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12490, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2999464273452759, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.13251947835087777, + "grad_norm": 0.0002733915753196925, + "kept_token_count": 572.7703125, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872025787830353, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.999013435735833e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2682769775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.684814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12500, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2986684501171112, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.1269302859902382, + "grad_norm": 0.0002495032676961273, + "kept_token_count": 574.0796875, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876062452793122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8413309559691695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.05719299316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7523956298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12510, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3000922560691833, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.12436038106679917, + "grad_norm": 0.0003531734982971102, + "kept_token_count": 572.6546875, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869872272014618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.083135984023102e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.18145141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7408386230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12520, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.29988168478012084, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.13343268930912017, + "grad_norm": 0.0003138607426080853, + "kept_token_count": 572.990625, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871506869792939, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.019275195081718e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0206634521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6354675292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12530, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.3003615587949753, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.13263184651732446, + "grad_norm": 0.00030881204293109477, + "kept_token_count": 572.4890625, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871110618114471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.034761998103932e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5170562744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9763702392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.30079677104949953, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.12690473198890687, + "grad_norm": 0.0004054211894981563, + "kept_token_count": 571.9953125, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871171832084655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0323667892371304e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9438140869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.82362365722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12550, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.30062510073184967, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.1324165292084217, + "grad_norm": 0.0002474871289450675, + "kept_token_count": 572.11875, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879335343837738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7134783744695594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.93212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.12449951171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12560, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2983097016811371, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.13173251897096633, + "grad_norm": 0.0001975942141143605, + "kept_token_count": 574.16875, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874692916870117, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.894832418358419e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.81915893554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.43712463378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12570, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3001088619232178, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.1369242325425148, + "grad_norm": 0.00023607676848769188, + "kept_token_count": 572.7609375, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883847534656525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.537221284408588e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.9138946533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0162048339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12580, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30129615366458895, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.13606654480099678, + "grad_norm": 0.00029095596983097494, + "kept_token_count": 571.7765625, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987585985660553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8492442147107795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.40862426757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.25164794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30066153407096863, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.12924075722694398, + "grad_norm": 0.0002804068208206445, + "kept_token_count": 572.3015625, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879319071769714, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.714115384558681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.35523376464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.39202270507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12600, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30040145516395567, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.13170330449938775, + "grad_norm": 0.0002769505954347551, + "kept_token_count": 572.3953125, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874855875968933, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8884649731917304e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8087677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2093170166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12610, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2995022892951965, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.13020637929439544, + "grad_norm": 0.0003814561932813376, + "kept_token_count": 573.1640625, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873670399188995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.934769858664367e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.6353332519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5342529296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12620, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.29974143505096434, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.12941595539450645, + "grad_norm": 0.00036374316550791264, + "kept_token_count": 572.8453125, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881300389766693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6367211689357644e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6850524902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0140716552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3000952899456024, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.13688726350665092, + "grad_norm": 0.00037652827450074255, + "kept_token_count": 572.56875, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874745070934295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.892790930171032e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.01958618164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6484893798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12640, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.29972404837608335, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.11722100228071212, + "grad_norm": 0.00025642989203333855, + "kept_token_count": 572.8890625, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877507805824279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7848735994193704e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.5758819580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44724731445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12650, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2999396830797195, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.13531836792826651, + "grad_norm": 0.0002878580999094993, + "kept_token_count": 572.8359375, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882851660251617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5761231012875216e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.13312072753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2257019042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2996673971414566, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.1245405375957489, + "grad_norm": 0.0001942441740538925, + "kept_token_count": 572.9640625, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875106394290925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8786797196953555e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6970275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.31220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12670, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30002325475215913, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.12772822827100755, + "grad_norm": 0.000229896220844239, + "kept_token_count": 573.0390625, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880391836166382, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.672216418839525e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.43540344238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9661926269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12680, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.30035406053066255, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.13300493285059928, + "grad_norm": 0.00023866572882980108, + "kept_token_count": 572.7578125, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881271839141845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.637832862499636e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.46502075195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.09222412109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12690, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.2989640265703201, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.12173451110720634, + "grad_norm": 0.00020103433053009212, + "kept_token_count": 573.5578125, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882060825824738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.607020091498271e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6546112060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1815948486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12700, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30062653720378874, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.13290340155363084, + "grad_norm": 0.0002867714501917362, + "kept_token_count": 572.3671875, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877917468547821, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.76886925753206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7066589355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01683654785154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12710, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.300320041179657, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.13490234240889548, + "grad_norm": 0.0002593683893792331, + "kept_token_count": 572.3859375, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878731310367584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7370747415698135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.24091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5023529052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12720, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.3009819179773331, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.13409795835614205, + "grad_norm": 0.0002544241724535823, + "kept_token_count": 571.875, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873613655567169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.936987643304747e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8429931640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1094024658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12730, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29994113445281984, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.12701603546738624, + "grad_norm": 0.00020255462732166052, + "kept_token_count": 572.8484375, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884702444076539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.50382889539469e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5002380371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.50538024902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12740, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.3001147240400314, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.13358161523938178, + "grad_norm": 0.00026484919362701476, + "kept_token_count": 572.565625, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881938517093658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6118009049678224e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2710296630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1781951904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12750, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30105866491794586, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.131870798766613, + "grad_norm": 0.0002538393600843847, + "kept_token_count": 572.0140625, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0, + "mgd_cos_sim": 0.98724285364151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.983282669854816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.28424072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.50368041992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12760, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2998416513204575, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.1351897969841957, + "grad_norm": 0.0002640250022523105, + "kept_token_count": 572.890625, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873739898204803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.932054689561482e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.3343139648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01341857910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12770, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.30086977779865265, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.13111638724803926, + "grad_norm": 0.000281724234810099, + "kept_token_count": 572.028125, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987553721666336, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8618453001836313e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.69224548339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9970428466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12780, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2997626900672913, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.14330884218215942, + "grad_norm": 0.0004994092741981149, + "kept_token_count": 572.853125, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986141049861908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.413664366642479e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0116668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.82993774414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12790, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30016557574272157, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.126922295242548, + "grad_norm": 0.0002324899542145431, + "kept_token_count": 572.4171875, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876068711280823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.841088921239134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.28358154296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.86040954589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12800, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.2998291730880737, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.1244947798550129, + "grad_norm": 0.00022664920834358782, + "kept_token_count": 572.9640625, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873940289020539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.924222557747271e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.77894897460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.21986389160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12810, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2996086686849594, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.13081480860710143, + "grad_norm": 0.0002480483672115952, + "kept_token_count": 573.3015625, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873345494270325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.947462111886125e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.24472961425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7507019042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12820, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.299590590596199, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.13135202303528787, + "grad_norm": 0.0003682725946418941, + "kept_token_count": 573.4, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9867941081523895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.158570274943486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.05128173828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6733825683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12830, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.29997937977313993, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.1251523956656456, + "grad_norm": 0.00022985345276538283, + "kept_token_count": 572.9140625, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865231931209564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.264397113933228e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.59075317382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.23121643066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12840, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30028326213359835, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.13553324937820435, + "grad_norm": 0.0002786148979794234, + "kept_token_count": 572.690625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986838448047638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1412531684036365e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0039978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6664245605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12850, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.29979413747787476, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.12676941156387328, + "grad_norm": 0.00024430110352113843, + "kept_token_count": 573.0234375, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875008940696717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.88247984321788e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.00409240722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1262603759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12860, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3000068336725235, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.1259428359568119, + "grad_norm": 0.0003066968056373298, + "kept_token_count": 572.6921875, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873860657215119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9273361946688964e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8752410888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3823638916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12870, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.299903017282486, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.1305037535727024, + "grad_norm": 0.0002420998935122043, + "kept_token_count": 572.7109375, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870872437953949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.04406270920299e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6935638427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.23583068847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12880, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.29962187111377714, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.1422957457602024, + "grad_norm": 0.0002370723959757015, + "kept_token_count": 572.9859375, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880333006381988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6745083818677816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.01625366210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0722991943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12890, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.3003047466278076, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.12985559403896332, + "grad_norm": 0.00023113949282560498, + "kept_token_count": 572.5265625, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881446719169616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.631007723219227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.85980224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.97138061523435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.29997076094150543, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.11589226573705673, + "grad_norm": 0.0003105121140833944, + "kept_token_count": 572.828125, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876474261283874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.825243777304422e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0807312011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3818328857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29897406995296477, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.13121672794222833, + "grad_norm": 0.0002364828687859699, + "kept_token_count": 573.603125, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988383537530899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.537699969660025e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.61670837402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.25095825195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12920, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2994616895914078, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.1389132559299469, + "grad_norm": 0.00023312093981076032, + "kept_token_count": 573.3171875, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.986930638551712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1052380513283424e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.4305725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.90549621582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12930, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.29973646700382234, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.1405879646539688, + "grad_norm": 0.00024443824077025056, + "kept_token_count": 573.10625, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875840663909912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8499961485504174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.5890289306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6864013671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12940, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.29947627782821656, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.12905253991484641, + "grad_norm": 0.0002861450775526464, + "kept_token_count": 573.4203125, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874065458774567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9193388258572666e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.663916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1212097167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12950, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.2997358053922653, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.12839925959706305, + "grad_norm": 0.00022624021221417934, + "kept_token_count": 572.9625, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868482708930969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1374143004068175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.272265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.68242797851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12960, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2998266011476517, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.1246388278901577, + "grad_norm": 0.00024161217152141035, + "kept_token_count": 572.73125, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877856850624085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.771237836393993e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.65106811523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9237426757812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12970, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.3003188967704773, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.1236997477710247, + "grad_norm": 0.00029084199923090637, + "kept_token_count": 572.6171875, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880424320697785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.670939124480355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.05096130371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9088623046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12980, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.30023767352104186, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.12601612955331803, + "grad_norm": 0.00023327668895944953, + "kept_token_count": 572.6265625, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878000915050507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.765608755405992e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0034576416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.73265380859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2998363584280014, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.1269723318517208, + "grad_norm": 0.0002496967790648341, + "kept_token_count": 572.86875, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9865757644176483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.243861087365076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.95115051269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9701416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13000, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2991022288799286, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.12821281254291533, + "grad_norm": 0.0003759065002668649, + "kept_token_count": 573.6046875, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872639358043671, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9750425387173894e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4235382080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.52293701171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13010, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.29976771473884584, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.12461879402399063, + "grad_norm": 0.0002595296537037939, + "kept_token_count": 573.0671875, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868733942508697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.1275927762617354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6867736816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5777618408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13020, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30013916492462156, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.12233806252479554, + "grad_norm": 0.0002463442215230316, + "kept_token_count": 572.6703125, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988096022605896, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.650009359465912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.0586791992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7651672363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13030, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2992303878068924, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.12287090718746185, + "grad_norm": 0.00024233698786702007, + "kept_token_count": 573.3359375, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870917797088623, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.042292723373975e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.6712677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7471588134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13040, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.3009188026189804, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.12685242518782616, + "grad_norm": 0.00030664101359434426, + "kept_token_count": 572.109375, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881214797496796, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.640064362320118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.066650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6712188720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13050, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.3002773433923721, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.1335857853293419, + "grad_norm": 0.00029531732434406877, + "kept_token_count": 572.5265625, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886095345020294, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.449427106010262e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.23499755859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4286315917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13060, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29917098581790924, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.1273520492017269, + "grad_norm": 0.0002384091931162402, + "kept_token_count": 573.65, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874469816684723, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.903540029772557e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.04125061035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.93873291015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13070, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29967889487743377, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.13531702384352684, + "grad_norm": 0.0002211432292824611, + "kept_token_count": 573.0984375, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872188568115234, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.992648791812826e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.23000183105466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2061706542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13080, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.3012953221797943, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.13503310680389405, + "grad_norm": 0.0003781951090786606, + "kept_token_count": 572.071875, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882967054843903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.571615863824263e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.90362243652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9599639892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13090, + "valid_token_count": 818.7703125 + }, + { + "actual_token_mask_ratio": 0.3006902515888214, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.12436759099364281, + "grad_norm": 0.00027943385066464543, + "kept_token_count": 572.178125, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878823161125183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7334946793853304e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.1364715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6501129150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13100, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29905246794223783, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.13884711787104606, + "grad_norm": 0.0002442820696160197, + "kept_token_count": 573.453125, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987324333190918, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9514517377247104e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.06605529785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.81275634765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13110, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.30009458065032957, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.13326094970107077, + "grad_norm": 0.00026006411644630134, + "kept_token_count": 572.8140625, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883744239807128, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.541260022961069e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9033447265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.64789123535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13120, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.2990132123231888, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.12608855441212655, + "grad_norm": 0.00028518031467683613, + "kept_token_count": 573.6921875, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872111916542053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.995649942429736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6028106689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1540832519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13130, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.300452920794487, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.12046233639121055, + "grad_norm": 0.0003157182363793254, + "kept_token_count": 572.4765625, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881725251674652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.620127547241282e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6896667480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.40001525878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.29890264570713043, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.12860240861773492, + "grad_norm": 0.0002687706146389246, + "kept_token_count": 573.653125, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987582802772522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.850481782341376e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.4239776611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.94521484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13150, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29974606931209563, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.1360067293047905, + "grad_norm": 0.0003611312713474035, + "kept_token_count": 573.125, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873976945877075, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.922804073430598e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.2509338378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6283905029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13160, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.29935502409935, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.13856832981109618, + "grad_norm": 0.00025364430621266365, + "kept_token_count": 573.421875, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870568692684174, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.055922738392837e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.57496337890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9895263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13170, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3004727065563202, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.13177455738186836, + "grad_norm": 0.0002866431896109134, + "kept_token_count": 572.2734375, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877115607261657, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8001869436120616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.25330810546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2201782226563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13180, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2991865277290344, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.13616034761071205, + "grad_norm": 0.0003364615549799055, + "kept_token_count": 573.365625, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9866757988929749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.204787375987507e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.23175354003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3232360839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13190, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2995454639196396, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.1262325420975685, + "grad_norm": 0.0002408192667644471, + "kept_token_count": 572.96875, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877955555915833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7673768494860266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.6341033935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.19024658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13200, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.29969391524791716, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.12984981387853622, + "grad_norm": 0.0002041120606008917, + "kept_token_count": 573.025, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880398154258728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.671974056691397e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2889434814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7806396484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13210, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3002126395702362, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.12772391512989997, + "grad_norm": 0.00023083461564965546, + "kept_token_count": 572.31875, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877407431602478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.788789738086052e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7623718261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.76629333496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13220, + "valid_token_count": 817.846875 + }, + { + "actual_token_mask_ratio": 0.2996107578277588, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.12997856512665748, + "grad_norm": 0.0001953871687874198, + "kept_token_count": 573.159375, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885212779045105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4838946996605956e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5341033935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.79088439941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13230, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30016453862190245, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.13292136192321777, + "grad_norm": 0.0002040291583398357, + "kept_token_count": 572.503125, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879283368587494, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.715513096016366e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.39215393066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7084228515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13240, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.30002107918262483, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.12503688633441926, + "grad_norm": 0.00023007362324278802, + "kept_token_count": 572.790625, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987970232963562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.699152232205961e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.0606201171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.85341186523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13250, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.29955007433891295, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.12751231268048285, + "grad_norm": 0.00021018389088567346, + "kept_token_count": 573.403125, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884049892425537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5293245057109745e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.57008666992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2572326660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13260, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.3001256287097931, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.12902901843190193, + "grad_norm": 0.0001972150057554245, + "kept_token_count": 572.2328125, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883307039737701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.558337532216683e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.88304748535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0543487548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13270, + "valid_token_count": 817.621875 + }, + { + "actual_token_mask_ratio": 0.3001314640045166, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.12382107824087143, + "grad_norm": 0.00027735793264582753, + "kept_token_count": 572.8625, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882184803485871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6021751768421384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6997009277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9348968505859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13280, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.30092352628707886, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.12619332149624823, + "grad_norm": 0.00027137575671076775, + "kept_token_count": 571.9875, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873166799545288, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9544450666871855e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.05384826660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6655303955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13290, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2991457343101501, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.13016136437654496, + "grad_norm": 0.00021226817625574768, + "kept_token_count": 573.3671875, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884347021579742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5177160427556376e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.50127563476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.17551879882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13300, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30036684572696687, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.1321467712521553, + "grad_norm": 0.0002613069664221257, + "kept_token_count": 572.3375, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9868907868862152, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.120805180922616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.363525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13310, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2999821871519089, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.1427215240895748, + "grad_norm": 0.00023893975594546646, + "kept_token_count": 572.98125, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878493010997772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.746389786305372e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.56480102539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.48333129882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13320, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.29975346326828, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.13105301782488823, + "grad_norm": 0.0002279983600601554, + "kept_token_count": 572.9984375, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876438558101654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8266423254972325e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.20897216796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.218408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13330, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2989907503128052, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.12238736003637314, + "grad_norm": 0.0002556316030677408, + "kept_token_count": 573.540625, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880683302879334, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6608279808424415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.23766174316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5122100830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13340, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.30056424140930177, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.13015293329954147, + "grad_norm": 0.0002725596132222563, + "kept_token_count": 572.0015625, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885400772094727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.476547510421369e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.2879272460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.36089477539065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13350, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.2999195158481598, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.13762337490916252, + "grad_norm": 0.0002803411625791341, + "kept_token_count": 572.9484375, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987823486328125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.756470916618127e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.119921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0755981445312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13360, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.3002677887678146, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.11593952402472496, + "grad_norm": 0.00028203619876876473, + "kept_token_count": 572.56875, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891413986682892, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.241662136337254e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.90468139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.30824890136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29997560381889343, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.13190924748778343, + "grad_norm": 0.00028448368539102376, + "kept_token_count": 573.1546875, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883537828922272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5493162542697976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.8511199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1322448730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13380, + "valid_token_count": 818.7625 + }, + { + "actual_token_mask_ratio": 0.3000978469848633, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.12081411629915237, + "grad_norm": 0.00032963475678116083, + "kept_token_count": 572.8390625, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878851294517517, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.732387205876876e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.17197875976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.71758728027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.3000584751367569, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.12906238809227943, + "grad_norm": 0.0003153171273879707, + "kept_token_count": 572.7234375, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873863637447358, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.927216832584236e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1508483886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7486083984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13400, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30071676075458526, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.12717680633068085, + "grad_norm": 0.0002913917996920645, + "kept_token_count": 572.321875, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.987122905254364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0301306328037756e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.66388549804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3094848632812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13410, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29942478239536285, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.1380024842917919, + "grad_norm": 0.00028658052906394005, + "kept_token_count": 573.3578125, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878624677658081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.741239572467748e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.03143920898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.42868957519534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13420, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2987110763788223, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.1353665016591549, + "grad_norm": 0.0002100467827403918, + "kept_token_count": 574.06875, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888215661048889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.366595640021842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.339501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9178100585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13430, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.29979003965854645, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.13354525193572045, + "grad_norm": 0.00027194374706596136, + "kept_token_count": 572.9875, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882394969463348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5939710253151135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.23680725097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7733093261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13440, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3009292811155319, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.12733163461089134, + "grad_norm": 0.0002529387129470706, + "kept_token_count": 572.053125, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880845487117768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6544896031264214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8104766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08091125488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13450, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.30001958906650544, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.12428709492087364, + "grad_norm": 0.00024004503211472183, + "kept_token_count": 572.6375, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888485610485077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3560519770835526e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3678863525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4049530029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13460, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2998759835958481, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.12912663519382478, + "grad_norm": 0.0002745452802628279, + "kept_token_count": 572.8640625, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874204695224762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9138983740704134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.775341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0267028808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13470, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29935763478279115, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.12448644563555718, + "grad_norm": 0.00020895774650853127, + "kept_token_count": 573.365625, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883412480354309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5542134830611755e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.47764587402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0752227783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2986127495765686, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.13031166270375252, + "grad_norm": 0.00023696910648141056, + "kept_token_count": 573.85, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880786776542664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6567825484089555e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.07667846679686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1371612548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2996963351964951, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.1308383122086525, + "grad_norm": 0.0003375689557287842, + "kept_token_count": 572.934375, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882916629314422, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5735888852505015e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.40302124023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.79676818847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13500, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.30029878616333006, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.13670426979660988, + "grad_norm": 0.00044066962436772883, + "kept_token_count": 572.415625, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870705008506775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.0506029947428035e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.34690551757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0301910400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13510, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.29951698184013364, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.12651813477277757, + "grad_norm": 0.0004016346065327525, + "kept_token_count": 573.1296875, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881838858127594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6156929965945895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2715240478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10303649902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13520, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.2994769006967545, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.12752382829785347, + "grad_norm": 0.00022652308689430356, + "kept_token_count": 573.2140625, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876283407211304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.832695922232233e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2462188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.19382934570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30064624547958374, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.12842243462800979, + "grad_norm": 0.00039285197271965444, + "kept_token_count": 572.034375, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883245289325714, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.560748493531719e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2598937988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.22643127441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13540, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.2995959550142288, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.12629079669713975, + "grad_norm": 0.00025752876535989344, + "kept_token_count": 573.1734375, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885556101799011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4704772881232203e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.9235015869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2701751708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13550, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3003419429063797, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.13487960025668144, + "grad_norm": 0.0002617499849293381, + "kept_token_count": 572.5703125, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872341096401215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9866994231706485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.16612854003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4207427978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13560, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.29907613396644595, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.13087198361754418, + "grad_norm": 0.0002153958921553567, + "kept_token_count": 573.5953125, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878608226776123, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.741880984511226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9318481445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5736999511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13570, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30085394382476804, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.1331300653517246, + "grad_norm": 0.0002273236314067617, + "kept_token_count": 572.1265625, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877789914608002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7738493958604525e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.23016052246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8484313964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13580, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.3005397289991379, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.12872890159487724, + "grad_norm": 0.0002264116337755695, + "kept_token_count": 572.1859375, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874614834785461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.897870749118738e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.34124755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.177294921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13590, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.2998164534568787, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.12533591985702514, + "grad_norm": 0.00027593132108449936, + "kept_token_count": 572.8171875, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880873799324036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6533912245649846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.83777770996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.03826293945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13600, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.3008877068758011, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.12444751039147377, + "grad_norm": 0.00022963092487771064, + "kept_token_count": 572.1734375, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873782157897949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.9304019194096325e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.53159790039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.16600646972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13610, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29891763627529144, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.12263977825641632, + "grad_norm": 0.00024678936460986733, + "kept_token_count": 573.5390625, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874604642391205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.898276529274881e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4163116455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.80714721679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13620, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.29983636140823366, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.12279843538999557, + "grad_norm": 0.0003243923420086503, + "kept_token_count": 573.1328125, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882032990455627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6081004256848246e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3800567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6102722167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13630, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30077630579471587, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.12591517344117165, + "grad_norm": 0.0002417794894427061, + "kept_token_count": 572.271875, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884176552295685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.524367250269279e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1698455810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7519287109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13640, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2995818257331848, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.12327322512865066, + "grad_norm": 0.0002827306161634624, + "kept_token_count": 573.209375, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876192927360534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8362361849285665e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.57631530761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9147583007813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13650, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30007438361644745, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.12243122085928917, + "grad_norm": 0.0002485199656803161, + "kept_token_count": 572.7765625, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888524115085602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3545411608647554e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0219451904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0100067138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2992806673049927, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.12863649502396585, + "grad_norm": 0.00019712245557457209, + "kept_token_count": 573.6109375, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873134016990661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.955719159625005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.8434143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5569580078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13670, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.2997113883495331, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.14036531895399093, + "grad_norm": 0.00020153823425062, + "kept_token_count": 572.90625, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877199530601501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.796908688149415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3914733886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0448425292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.3004171669483185, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.12852853760123253, + "grad_norm": 0.00026568686007522047, + "kept_token_count": 572.4578125, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869979977607727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.078926515125204e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.9766845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0952117919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13690, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.30025782585144045, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.13643087968230247, + "grad_norm": 0.0002961840946227312, + "kept_token_count": 572.49375, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880283772945404, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.676432909036521e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.22421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2274993896484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13700, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.30031501352787016, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.13038231804966927, + "grad_norm": 0.0002866133872885257, + "kept_token_count": 572.3875, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880918085575103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6516549264197235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.93578186035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.56549072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.3000387609004974, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.11949458047747612, + "grad_norm": 0.0003784284635912627, + "kept_token_count": 573.0125, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876842617988586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.810856480617076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.36936340332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9013427734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.29985193312168124, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.12344226688146591, + "grad_norm": 0.00024613519781269133, + "kept_token_count": 572.671875, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877788066864014, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.773919536091853e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.7139068603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7717712402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13730, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.3005686581134796, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.11967144161462784, + "grad_norm": 0.00023252135724760592, + "kept_token_count": 572.54375, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887977182865143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.37591248555691e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.08855895996095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4284637451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13740, + "valid_token_count": 818.59375 + }, + { + "actual_token_mask_ratio": 0.29931447505950926, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.1355590581893921, + "grad_norm": 0.00037060026079416275, + "kept_token_count": 573.4203125, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884868323802948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4973539479542525e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.02427673339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7422210693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13750, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.29995201230049134, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.13880974501371385, + "grad_norm": 0.0002738857874646783, + "kept_token_count": 572.7890625, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870077252388001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.075126937299501e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.80931396484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5076477050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.30029498040676117, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.12196643650531769, + "grad_norm": 0.0002344682434340939, + "kept_token_count": 572.9109375, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884908497333527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.495777502597775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9349639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4762756347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13770, + "valid_token_count": 818.7765625 + }, + { + "actual_token_mask_ratio": 0.3007971227169037, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.13349808901548385, + "grad_norm": 0.0002529587072785944, + "kept_token_count": 572.203125, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881606519222259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.624767316272482e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6024871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.85898742675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29948061108589175, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.12964724525809287, + "grad_norm": 0.0002427994622848928, + "kept_token_count": 573.0265625, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884591400623322, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.508166130108293e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6791717529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3180419921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13790, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.29999945163726804, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.12756117954850196, + "grad_norm": 0.0002649759117048234, + "kept_token_count": 572.696875, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881683886051178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.621737607521936e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3769104003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1850219726563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2995615154504776, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.12736741080880165, + "grad_norm": 0.00023448617139365524, + "kept_token_count": 573.3015625, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878099918365478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7617393647669815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.13214111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3504913330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13810, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.3005773276090622, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.12834876030683517, + "grad_norm": 0.00030392903136089444, + "kept_token_count": 572.2171875, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9869138479232789, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.111798418511171e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.35569152832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0557830810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13820, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.3002989649772644, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.12827517613768577, + "grad_norm": 0.00045316354953683913, + "kept_token_count": 572.621875, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877250373363495, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7949271538527684e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0973388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.24558715820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13830, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29968251287937164, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.13666188791394235, + "grad_norm": 0.00021590705728158355, + "kept_token_count": 573.171875, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882543802261352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5881560072302816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1080383300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8982269287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13840, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.30040477514266967, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.126438856869936, + "grad_norm": 0.00027978638536296785, + "kept_token_count": 572.3109375, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0, + "mgd_cos_sim": 0.987582677602768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.85053602460539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0023651123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8789489746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13850, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.29935320615768435, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.11788818761706352, + "grad_norm": 0.00022576912306249142, + "kept_token_count": 573.4703125, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988586962223053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4582363625522706e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7885223388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42982482910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13860, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.299693238735199, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.12993980795145035, + "grad_norm": 0.00019656098447740078, + "kept_token_count": 573.1578125, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877235531806946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7955055197235194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2422637939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.10814514160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13870, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.3001507371664047, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.13225919753313065, + "grad_norm": 0.00020657684945035726, + "kept_token_count": 572.8171875, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881629168987274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.623878412530757e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.3609191894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8631652832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13880, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.29906025528907776, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.13419344499707223, + "grad_norm": 0.00030342015088535845, + "kept_token_count": 573.3046875, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878850996494293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.732402048830408e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.7253356933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.55302734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13890, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.29933360815048216, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.12889784574508667, + "grad_norm": 0.0002418019575998187, + "kept_token_count": 573.315625, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873101055622101, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.957007658958901e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.78638610839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9452850341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13900, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.2983339250087738, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.13716360330581664, + "grad_norm": 0.0002676474105101079, + "kept_token_count": 574.025, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889389932155609, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.32072913099546e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.78912963867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.17347412109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3011350601911545, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.12698855325579644, + "grad_norm": 0.00026142565184272826, + "kept_token_count": 571.71875, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879657447338104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.700903336924966e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5177856445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2955596923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13920, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3006079375743866, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.12586765512824058, + "grad_norm": 0.00020616764959413558, + "kept_token_count": 572.1265625, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892174422740936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2119576755794695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.83995056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3394744873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13930, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.3007130265235901, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.11757812201976776, + "grad_norm": 0.0002833533799275756, + "kept_token_count": 572.38125, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877293348312378, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.793254665855784e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.15300903320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2466979980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13940, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.30037752389907835, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.12942871749401091, + "grad_norm": 0.00021752502652816474, + "kept_token_count": 572.371875, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880711734294891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.659713522414677e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.68760070800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4983856201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13950, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29974196255207064, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.13890246599912642, + "grad_norm": 0.00023295961727853864, + "kept_token_count": 573.08125, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888856530189514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.341561507317238e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3182800292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.99969177246095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13960, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.29927629232406616, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.13783570900559425, + "grad_norm": 0.00030992639949545264, + "kept_token_count": 573.184375, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883439540863037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.553168146230746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6136169433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7102386474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13970, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.2998077929019928, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.13438856452703477, + "grad_norm": 0.00023254849656950682, + "kept_token_count": 572.678125, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886945366859436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.416216324898414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9568389892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5735748291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13980, + "valid_token_count": 817.8828125 + }, + { + "actual_token_mask_ratio": 0.2999919533729553, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.12418785020709038, + "grad_norm": 0.00019651064940262586, + "kept_token_count": 572.6109375, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875962257385253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8452438568347135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.55260314941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2433685302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13990, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.30073348581790926, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.12344991788268089, + "grad_norm": 0.00030576129211112857, + "kept_token_count": 572.2734375, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879756331443786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.697035146818962e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6803680419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89690551757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14000, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2995857268571854, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.1315096102654934, + "grad_norm": 0.00022970011923462152, + "kept_token_count": 572.7375, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880980908870697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6492026376654395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4622100830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6582794189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14010, + "valid_token_count": 817.7140625 + }, + { + "actual_token_mask_ratio": 0.3000016719102859, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.12276033535599709, + "grad_norm": 0.0002075307274935767, + "kept_token_count": 572.7921875, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882322311401367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.596802064043004e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5603546142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.76918334960936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14020, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.29949587881565093, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.12068267837166786, + "grad_norm": 0.0002382322563789785, + "kept_token_count": 573.090625, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884825468063354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.499027236306574e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0003234863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.03062133789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14030, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3011076062917709, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.11921995356678963, + "grad_norm": 0.00021639031183440238, + "kept_token_count": 571.875, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883864283561706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.536569249466993e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.1266845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.29107666015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14040, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.3004496663808823, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.13025856539607047, + "grad_norm": 0.0002185021439800039, + "kept_token_count": 572.525, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884222328662873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.522586568782572e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.38360900878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7357604980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14050, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.29968178570270537, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.12744635641574859, + "grad_norm": 0.00022241217084228992, + "kept_token_count": 572.9484375, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873642086982727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.935868746542837e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.7964233398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1809509277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14060, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.29955500066280366, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.11995067819952965, + "grad_norm": 0.00025756657123565674, + "kept_token_count": 573.553125, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879980683326721, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.688268600148149e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.39361572265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.56357727050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14070, + "valid_token_count": 818.8453125 + }, + { + "actual_token_mask_ratio": 0.2998756647109985, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.126649821549654, + "grad_norm": 0.0003082373004872352, + "kept_token_count": 572.8046875, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873303890228271, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.949089634465054e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9365936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43793029785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14080, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.299629744887352, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.1277047574520111, + "grad_norm": 0.0002473039785400033, + "kept_token_count": 572.81875, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870995163917542, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.03927261888748e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7545654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14090, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.3006974637508392, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.12644784897565842, + "grad_norm": 0.0002857160579878837, + "kept_token_count": 572.2640625, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879404485225678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.710784778581001e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.8761474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.32379150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14100, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2997111976146698, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.12084075585007667, + "grad_norm": 0.00028016886790283024, + "kept_token_count": 572.753125, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885013222694397, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4916908154846166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.94683532714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.31968994140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14110, + "valid_token_count": 817.8859375 + }, + { + "actual_token_mask_ratio": 0.2990504950284958, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.12321504428982735, + "grad_norm": 0.0002545975148677826, + "kept_token_count": 573.6125, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880342185497284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.67415178718511e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2519073486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8928680419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14120, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30067365169525145, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.12762190476059915, + "grad_norm": 0.00021729069703724235, + "kept_token_count": 572.2234375, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882761240005493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.579660853778478e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3080657958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5830139160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14130, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2985324651002884, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.13138344883918762, + "grad_norm": 0.00020853769092354923, + "kept_token_count": 574.0109375, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890670239925384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.270716090104543e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1569030761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43039855957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14140, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3010711669921875, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.12761434391140938, + "grad_norm": 0.0003631350409705192, + "kept_token_count": 572.1703125, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888954997062683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.337718055467121e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.30311889648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8356872558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14150, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.30010633170604706, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.12488981112837791, + "grad_norm": 0.0003312089538667351, + "kept_token_count": 572.6625, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875229477882386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.873871803283692e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3786285400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6636260986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14160, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.29958591759204867, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.12897022739052771, + "grad_norm": 0.0002965005405712873, + "kept_token_count": 573.1859375, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889245569705963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.326363668951672e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3525024414063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.97997131347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3002342790365219, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.12468218952417373, + "grad_norm": 0.00024465913884341717, + "kept_token_count": 572.609375, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988248074054718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.590616554196458e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.9196716308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9745758056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14180, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.299320325255394, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.12716005966067315, + "grad_norm": 0.0002798985515255481, + "kept_token_count": 573.3234375, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876648247241974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.81844679597998e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4582763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2090698242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.30049778819084166, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.12862331718206405, + "grad_norm": 0.0002359342761337757, + "kept_token_count": 572.328125, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882678389549255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5828992369933985e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.3960815429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1384307861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30044168829917905, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.12279702350497246, + "grad_norm": 0.0002646806533448398, + "kept_token_count": 572.5453125, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881517231464386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.628247224900406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9199554443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.8798583984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14210, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.30043433904647826, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.13159106075763702, + "grad_norm": 0.00024287402629852295, + "kept_token_count": 572.2390625, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878946423530579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.728672283818014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.29969482421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9592620849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14220, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.2993825554847717, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.13933716639876365, + "grad_norm": 0.00041364607750438154, + "kept_token_count": 573.3625, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882041811943054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6077560909907334e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.29725646972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.03379516601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14230, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.30002314746379855, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.13417157605290414, + "grad_norm": 0.00029489409644156694, + "kept_token_count": 572.5234375, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885315239429474, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4798913586419074e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.92921447753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8544128417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14240, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.29915910959243774, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.1357632890343666, + "grad_norm": 0.00024296087212860584, + "kept_token_count": 573.284375, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879552900791169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7049917702679524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.47496032714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.98097839355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29971167743206023, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.12863880768418312, + "grad_norm": 0.00026668456848710775, + "kept_token_count": 573.1140625, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880685567855835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6607386320829394e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.45859680175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1215484619141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14260, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2996321201324463, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.13013385385274887, + "grad_norm": 0.0002748789556790143, + "kept_token_count": 573.021875, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885229289531707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.48324870376382e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4772247314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2651123046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14270, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.3006307065486908, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.12834391072392465, + "grad_norm": 0.00024593426496721804, + "kept_token_count": 572.4484375, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9871615052223206, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.015061833546497e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0697082519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.28587646484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14280, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.29973980188369753, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.12987937331199645, + "grad_norm": 0.0002176609996240586, + "kept_token_count": 573.0546875, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885436832904816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.475137793633621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9049560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.81505126953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2994664549827576, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.1328490823507309, + "grad_norm": 0.0002507233584765345, + "kept_token_count": 573.2203125, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881660282611847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.622662963811308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4541290283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2092742919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14300, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30077255368232725, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.13203420788049697, + "grad_norm": 0.00022972063743509352, + "kept_token_count": 572.053125, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879577994346619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7040033314260656e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2395477294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.47252502441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14310, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30049129128456115, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.1312360018491745, + "grad_norm": 0.000195647357031703, + "kept_token_count": 572.5359375, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886159479618073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4469157728599384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.08408508300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6200317382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14320, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.3002512902021408, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.12983420342206956, + "grad_norm": 0.00035814326838590205, + "kept_token_count": 572.6203125, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0001, + "mgd_cos_sim": 0.9870130836963653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 5.073031898064073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.73517456054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5839416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14330, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.29952854216098784, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.14073114693164826, + "grad_norm": 0.00026598849217407405, + "kept_token_count": 572.9125, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886924922466278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4170127512188626e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.2390594482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.18475341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14340, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.3013073593378067, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.12464992254972458, + "grad_norm": 0.00020118639804422855, + "kept_token_count": 571.6796875, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887583971023559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3912687397096305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.5308044433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3566101074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.3002527952194214, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.12916849479079245, + "grad_norm": 0.0002639952872414142, + "kept_token_count": 572.5109375, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884052395820617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5292249706108126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.44530334472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6875823974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14360, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29902470111846924, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.12172156199812889, + "grad_norm": 0.00022015794820617884, + "kept_token_count": 573.8609375, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878551185131073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7441154674743305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.36602478027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72630310058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14370, + "valid_token_count": 818.6671875 + }, + { + "actual_token_mask_ratio": 0.3007792204618454, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.12108588814735413, + "grad_norm": 0.00019784354662988335, + "kept_token_count": 572.365625, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887732088565826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3854863906744865e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8144104003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.69541931152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14380, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3003989666700363, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.13140763565897942, + "grad_norm": 0.0002524338196963072, + "kept_token_count": 572.325, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878377258777619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.750907864945475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.10035095214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2774139404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14390, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.29978094100952146, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.12550905868411064, + "grad_norm": 0.0003009759820997715, + "kept_token_count": 572.915625, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887019991874695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.413301394379232e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8398864746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3976989746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14400, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2993423819541931, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.1328126758337021, + "grad_norm": 0.0002768437552731484, + "kept_token_count": 573.2390625, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883617281913757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5462195339496246e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5421478271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0456146240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14410, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3005222141742706, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.1192897729575634, + "grad_norm": 0.0002205173805123195, + "kept_token_count": 572.2375, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881621956825256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6241665768320675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.341357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9036865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14420, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2996314465999603, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.13434874042868614, + "grad_norm": 0.0003413173835724592, + "kept_token_count": 572.9109375, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875080943107605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.879667903878726e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8481811523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6288269042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14430, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.30051020085811614, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.12596103250980378, + "grad_norm": 0.00020858939387835562, + "kept_token_count": 572.346875, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890752792358398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.267484000592958e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3558624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.56055908203126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14440, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3004173457622528, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.12732992693781853, + "grad_norm": 0.00027905020397156477, + "kept_token_count": 572.534375, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887759685516357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3844013634952714e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7029602050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3232025146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14450, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3022498697042465, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.11932322531938552, + "grad_norm": 0.00020641366427298635, + "kept_token_count": 570.665625, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891209542751312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.249653211445548e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2313232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5213287353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14460, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.29952738881111146, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.1284467615187168, + "grad_norm": 0.00033936696127057076, + "kept_token_count": 573.3359375, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879950284957886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.689457309723366e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.36468200683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2967254638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14470, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.300571608543396, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.12695179283618926, + "grad_norm": 0.00020846747793257236, + "kept_token_count": 572.534375, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886370658874511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.438666946953163e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8207153320312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6177062988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14480, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.2999365717172623, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.13334725201129913, + "grad_norm": 0.00021404793369583786, + "kept_token_count": 572.7859375, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9873907625675201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.92550552735338e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.7950012207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2748565673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14490, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.29994718730449677, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.12176450788974762, + "grad_norm": 0.00019806875206995755, + "kept_token_count": 572.984375, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883196115493774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.562675967463292e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0314239501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.45990295410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14500, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.2998252213001251, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.13522423058748245, + "grad_norm": 0.0002271657867822796, + "kept_token_count": 572.946875, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878583192825318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7428602556465194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.18214111328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5583984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14510, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2985850512981415, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.13019824475049974, + "grad_norm": 0.00025982552324421704, + "kept_token_count": 574.1140625, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881564557552338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.62640258774627e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.4355194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.37708435058596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14520, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.3003072261810303, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.1322281278669834, + "grad_norm": 0.00023069938470143825, + "kept_token_count": 572.21875, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881557345390319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.626688023563474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7569519042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.40321960449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14530, + "valid_token_count": 817.80625 + }, + { + "actual_token_mask_ratio": 0.3008433163166046, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.12821297943592072, + "grad_norm": 0.00024342596589121968, + "kept_token_count": 572.125, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888744592666626, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.345927591202781e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2550750732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5593963623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14540, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.30016026496887205, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.1411172054708004, + "grad_norm": 0.0002608762588351965, + "kept_token_count": 572.6453125, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880098342895508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.68367576104356e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.64837341308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3538848876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14550, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3012765318155289, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.13027825206518173, + "grad_norm": 0.00025517199537716806, + "kept_token_count": 571.6421875, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988404506444931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.529509860731196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.39708251953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.58031311035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14560, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3001199573278427, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.12184449434280395, + "grad_norm": 0.00020177761325612664, + "kept_token_count": 572.6796875, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883723139762879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.542090609902516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1379028320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.83429260253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14570, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.2994798064231873, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.12270125150680541, + "grad_norm": 0.00026710485690273345, + "kept_token_count": 573.19375, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878435790538788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.748619030578993e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.60216064453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65285034179686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14580, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2991914749145508, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.12906923666596412, + "grad_norm": 0.0003190077841281891, + "kept_token_count": 573.4375, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880265474319458, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6771447523497044e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8575897216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.56313171386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14590, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.299869692325592, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.11917324587702752, + "grad_norm": 0.0002178251597797498, + "kept_token_count": 572.871875, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894714891910553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.11271263146773e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.938916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3978515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14600, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3014122128486633, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.13141146302223206, + "grad_norm": 0.0002835956693161279, + "kept_token_count": 571.4484375, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879746079444885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6974374708952385e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8115966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36809692382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14610, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.3006537199020386, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.12467742413282394, + "grad_norm": 0.00020607213082257658, + "kept_token_count": 572.1296875, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882629990577698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5847829824197106e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2127746582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9750946044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14620, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.29989021420478823, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.11552837565541267, + "grad_norm": 0.00021584190835710615, + "kept_token_count": 572.915625, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886123180389405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4483362216851674e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6470001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.906494140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.29962819516658784, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.1357191286981106, + "grad_norm": 0.00023467755818273872, + "kept_token_count": 573.0484375, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891779899597168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.227370591252111e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.4637481689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.346533203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14640, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2993812680244446, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.12641641944646836, + "grad_norm": 0.0002721514902077615, + "kept_token_count": 573.3625, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876203060150146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.835838517465163e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.06548767089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28705444335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29987357556819916, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.12800448313355445, + "grad_norm": 0.0002582167217042297, + "kept_token_count": 572.934375, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879786312580109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.69586557301227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.92552490234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.11427612304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14660, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29807342290878297, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.13676470667123794, + "grad_norm": 0.00023018168576527387, + "kept_token_count": 574.45625, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877366721630096, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.790378952748142e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8696350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.74391479492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14670, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2998321622610092, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.12736227735877037, + "grad_norm": 0.00025328679475933313, + "kept_token_count": 573.1265625, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879076778888702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.723584170278628e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.88807373046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3487518310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14680, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.30077713429927827, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.12206944897770881, + "grad_norm": 0.00022286063176579773, + "kept_token_count": 572.109375, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888964354991913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.337345235398971e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.51271362304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.69847412109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14690, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2998648166656494, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.13501040190458297, + "grad_norm": 0.00028249662136659026, + "kept_token_count": 573.0078125, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886422872543335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.436628405528609e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.535693359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2028839111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14700, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.301018488407135, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.12773109450936318, + "grad_norm": 0.0002452127810101956, + "kept_token_count": 572.1140625, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882465064525604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.591225770127494e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2517028808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6252868652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14710, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.299213120341301, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.13437072858214377, + "grad_norm": 0.0002198397705797106, + "kept_token_count": 573.765625, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988065880537033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.661783495976124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.31020202636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0987121582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14720, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.3007804095745087, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.1312215246260166, + "grad_norm": 0.000282899709418416, + "kept_token_count": 572.2328125, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887666702270508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3880425437237136e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.07924194335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6868011474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14730, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30026909410953523, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.12033810690045357, + "grad_norm": 0.0002955990785267204, + "kept_token_count": 572.3828125, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885635614395142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.467379221750889e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.12782897949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6528594970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14740, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2993452399969101, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.12645122334361075, + "grad_norm": 0.0002200002345489338, + "kept_token_count": 573.1703125, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885768592357635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4621843335335144e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0879608154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0107177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14750, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.3003863513469696, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.12408882454037666, + "grad_norm": 0.0001968130236491561, + "kept_token_count": 572.8, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887001812458038, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.414005743456073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.361865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2169677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14760, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.30065923631191255, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.12935834154486656, + "grad_norm": 0.0002161585580324754, + "kept_token_count": 572.55625, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890974402427674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2588337964843956e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0506896972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.843701171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14770, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.2992743581533432, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.1323030784726143, + "grad_norm": 0.00019031537522096187, + "kept_token_count": 573.3359375, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880923092365265, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6514618225046435e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2716094970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.62945556640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14780, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.29880974590778353, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.12110757455229759, + "grad_norm": 0.00022523840016219765, + "kept_token_count": 573.7578125, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881126999855041, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.643495194613934e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.9118194580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.19158935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14790, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29842716455459595, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.12928965613245963, + "grad_norm": 0.00020405344548635185, + "kept_token_count": 573.9734375, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886901021003723, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.417951931827702e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.46905517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.60556945800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14800, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.300705274939537, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.13474305495619773, + "grad_norm": 0.00027351893368177116, + "kept_token_count": 572.440625, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883504450321198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.550627199932933e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.11217346191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1580108642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14810, + "valid_token_count": 818.5984375 + }, + { + "actual_token_mask_ratio": 0.3008701533079147, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.1275571532547474, + "grad_norm": 0.00020032429893035442, + "kept_token_count": 572.2328125, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890429615974426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.280110806575976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.84803161621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4461334228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14820, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.2997495889663696, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.1293174706399441, + "grad_norm": 0.00029976238147355616, + "kept_token_count": 573.0984375, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9872598469257354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.976644086127635e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.22905883789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4020111083984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14830, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.3002763450145721, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.1279424861073494, + "grad_norm": 0.0002844833361450583, + "kept_token_count": 572.4109375, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882364451885224, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5951562060508874e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1766815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.54127502441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14840, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.29996767044067385, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.11129733547568321, + "grad_norm": 0.00021126610226929188, + "kept_token_count": 572.8046875, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881944775581359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.611553995346185e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9635772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.96583862304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14850, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30101464688777924, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.12707732394337654, + "grad_norm": 0.00023371263523586094, + "kept_token_count": 572.053125, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889143526554107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3303486381773834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8122497558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.54053955078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14860, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29969754815101624, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.12727108746767044, + "grad_norm": 0.0002725040540099144, + "kept_token_count": 573.071875, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882205963134766, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.601351720339153e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1647064208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.59527893066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14870, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2986018300056458, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.12903562933206558, + "grad_norm": 0.0002582808374427259, + "kept_token_count": 573.89375, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988491702079773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.495445791690145e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7607482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.52577514648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14880, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.300059500336647, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.12255652770400047, + "grad_norm": 0.00017397929332219064, + "kept_token_count": 572.9, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988861757516861, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.350897652329877e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.379052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1656890869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14890, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.29952540397644045, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.13274437114596366, + "grad_norm": 0.00040263973642140627, + "kept_token_count": 573.3453125, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883782863616943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.539751353149768e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.63116455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6418731689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14900, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.300060772895813, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.12831235602498053, + "grad_norm": 0.00021268072305247188, + "kept_token_count": 572.6296875, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884872078895569, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.497205918596592e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3457336425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.28916320800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30029901266098025, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.12140262871980667, + "grad_norm": 0.00027187331579625607, + "kept_token_count": 572.528125, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884813904762269, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4994736526859924e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2786590576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.69237365722654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14920, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29982914924621584, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.1402814842760563, + "grad_norm": 0.00025718475808389485, + "kept_token_count": 572.9375, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9878978490829468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.727421874122229e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.27137756347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7256713867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14930, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.30034577250480654, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.12745126336812973, + "grad_norm": 0.00035482915700413287, + "kept_token_count": 572.5890625, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885087728500366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.488780577958096e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.92340698242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.21111450195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14940, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29933472275733947, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.12155078500509262, + "grad_norm": 0.00024623775971122086, + "kept_token_count": 573.35, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880163609981537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.681124773924239e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.78801879882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2515899658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14950, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2995890974998474, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.12191702425479889, + "grad_norm": 0.00017517399101052433, + "kept_token_count": 573.3140625, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988898766040802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.336435449658893e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.41969604492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1255676269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14960, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29963316321372985, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.13128756955266, + "grad_norm": 0.00028252220363356173, + "kept_token_count": 573.4296875, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877158880233765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.798499794560485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7336486816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7862609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14970, + "valid_token_count": 818.7609375 + }, + { + "actual_token_mask_ratio": 0.3005411624908447, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.128396362811327, + "grad_norm": 0.0001708358176983893, + "kept_token_count": 572.171875, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887169897556305, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.407441592775285e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0574432373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.00145568847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14980, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.29999114871025084, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.12722970470786094, + "grad_norm": 0.00021738838404417038, + "kept_token_count": 572.75625, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891690194606781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.230873346386943e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1570159912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.578564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14990, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2996083199977875, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.12494276314973832, + "grad_norm": 0.00019443369819782674, + "kept_token_count": 573.190625, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888189375400543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3676235145539975e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.6138458251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.84393310546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15000, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2983189344406128, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.13078014254570008, + "grad_norm": 0.00027102945023216307, + "kept_token_count": 574.321875, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882505357265472, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.589652417053003e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.63592834472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2004730224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15010, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.3004879981279373, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.12952386811375619, + "grad_norm": 0.0002396304189460352, + "kept_token_count": 572.4671875, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885428547859192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.475466957956087e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1656005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.25812072753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15020, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2999593198299408, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.11621607020497322, + "grad_norm": 0.0001955026382347569, + "kept_token_count": 572.853125, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883740305900574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.541422349575441e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.10667114257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5948486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15030, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.30011523962020875, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.12549634426832199, + "grad_norm": 0.00030365987913683057, + "kept_token_count": 572.7609375, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885941028594971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.455446687643416e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5809692382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.77417907714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2992513984441757, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.12607263699173926, + "grad_norm": 0.0003654677129816264, + "kept_token_count": 573.51875, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892111241817474, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.214429427520372e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9671112060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0932342529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15050, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.29982729256153107, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.12802156880497934, + "grad_norm": 0.00026324368081986904, + "kept_token_count": 573.046875, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898484528064728, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9654627471463755e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.95631408691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4947113037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15060, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2996918439865112, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.13853791281580924, + "grad_norm": 0.00026973988860845566, + "kept_token_count": 572.8859375, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887234032154083, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.40493764472194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7442321777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61692199707034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15070, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.30145685076713563, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.1373388111591339, + "grad_norm": 0.0002476571244187653, + "kept_token_count": 571.678125, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886084020137786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.449860207387246e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.3806457519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7371063232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15080, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.3005793750286102, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.12566139698028564, + "grad_norm": 0.00024452913203276694, + "kept_token_count": 572.21875, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885617315769195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.468090082809795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1730072021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7710906982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15090, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2992702692747116, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.12254865393042565, + "grad_norm": 0.00029788765823468566, + "kept_token_count": 573.2890625, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886898517608642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.418046410137322e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1090362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.01793212890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15100, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.30063164234161377, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.13481439724564553, + "grad_norm": 0.0002217187429778278, + "kept_token_count": 572.096875, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882317066192627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5970075734658164e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.9446533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1512756347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15110, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.30005630254745486, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.12264624312520027, + "grad_norm": 0.00026077788788825274, + "kept_token_count": 572.9046875, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892134487628936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.213518841424957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5918426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3631927490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15120, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2997607707977295, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.1214309923350811, + "grad_norm": 0.0001958946231752634, + "kept_token_count": 573.18125, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882231593132019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.600346837833058e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4013702392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2836883544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15130, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.3002090692520142, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.12833559960126878, + "grad_norm": 0.0002771560684777796, + "kept_token_count": 572.76875, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877695620059967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.777534704771824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.36301574707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.41917724609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15140, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3007250964641571, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.11685003116726875, + "grad_norm": 0.00018290035950485617, + "kept_token_count": 572.0140625, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888721406459808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3468423245940355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.37779235839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1357666015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15150, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.300795903801918, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.12062573209404945, + "grad_norm": 0.0003637201734818518, + "kept_token_count": 571.9203125, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879644393920899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.7014118899824096e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.79017333984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8596557617187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15160, + "valid_token_count": 817.965625 + }, + { + "actual_token_mask_ratio": 0.30108789205551145, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.137524501234293, + "grad_norm": 0.0002481494448147714, + "kept_token_count": 571.7703125, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882075428962708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.606440852512605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6560455322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0252655029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15170, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3001769095659256, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.1278531312942505, + "grad_norm": 0.0002222847833763808, + "kept_token_count": 572.7890625, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882803678512573, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5779936408507636e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4999481201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17710876464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15180, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.29992373287677765, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.13989573940634728, + "grad_norm": 0.00022930969134904444, + "kept_token_count": 572.825, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879749059677124, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6973272401373836e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.4583404541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06162414550784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.3002964943647385, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.12948875948786737, + "grad_norm": 0.00020029173174407333, + "kept_token_count": 572.3984375, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887487411499023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.395037431095261e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.45018005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9974853515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15200, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.3008462816476822, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.13305791094899178, + "grad_norm": 0.0002244484785478562, + "kept_token_count": 572.2265625, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888635337352752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3502036714926365e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.03692932128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0713195800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15210, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.30028554499149324, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.13633189722895622, + "grad_norm": 0.00024146986834239215, + "kept_token_count": 572.359375, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886513352394104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4330896344035864e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9956085205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5498291015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15220, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29958059787750246, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.1325788192451, + "grad_norm": 0.00016956038598436862, + "kept_token_count": 573.1359375, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889095246791839, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.332234238972888e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3962127685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.52478942871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15230, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.2984238415956497, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.12765747234225272, + "grad_norm": 0.0002886993170250207, + "kept_token_count": 574.1640625, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881233215332031, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.639344842871651e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.55767517089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43348083496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.299841383099556, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.121159016340971, + "grad_norm": 0.0002219049638370052, + "kept_token_count": 572.74375, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886210203170777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.444937039806973e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6424560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42943115234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15250, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.29887695610523224, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.11079800203442573, + "grad_norm": 0.00025808362988755107, + "kept_token_count": 573.625, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885286152362823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.481031173781957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.44492797851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.15513610839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15260, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.30015005767345426, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.13408379033207893, + "grad_norm": 0.00019891612464562058, + "kept_token_count": 572.61875, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880765616893769, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6576094609918074e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.60125427246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1627960205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15270, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.30031132996082305, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.12803444489836693, + "grad_norm": 0.0002530495694372803, + "kept_token_count": 572.6140625, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886485576629639, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.434174297784921e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9266296386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6209411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15280, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3006760150194168, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.12158307805657387, + "grad_norm": 0.00023059037630446255, + "kept_token_count": 572.19375, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890193104743957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.289350763428956e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.34063415527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.54725341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15290, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.3001001447439194, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.12857668548822404, + "grad_norm": 0.00025445970823056996, + "kept_token_count": 572.7171875, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881888806819916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.613738929037936e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.83576049804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8828186035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15300, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.2990859806537628, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.12358122020959854, + "grad_norm": 0.00024309205764438957, + "kept_token_count": 573.5546875, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875380635261536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.867960378760472e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2201171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.00035705566404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15310, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30152284204959867, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.12044828459620475, + "grad_norm": 0.00020283913181629032, + "kept_token_count": 571.6046875, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9877117097377777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8001283357734795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6216339111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0369049072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15320, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30069861114025115, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.12888620495796205, + "grad_norm": 0.000248355558142066, + "kept_token_count": 572.0078125, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888932585716248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.338590078987181e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.9899963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.47999267578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15330, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.29985814690589907, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.12797950953245163, + "grad_norm": 0.00025693688075989485, + "kept_token_count": 572.89375, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989531260728836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.089375943294726e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6933654785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8331298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15340, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30056631565093994, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.13261353150010108, + "grad_norm": 0.00020247124484740198, + "kept_token_count": 572.2484375, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889066755771637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.333346078055911e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.43329467773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7504058837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15350, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2997312694787979, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.1326556034386158, + "grad_norm": 0.0002528756158426404, + "kept_token_count": 573.215625, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892518758773804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.198499591439031e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.87579956054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.73541870117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15360, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.29975707530975343, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.13590000495314597, + "grad_norm": 0.0003456226841080934, + "kept_token_count": 573.06875, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885168552398682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.485619174374733e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8942169189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.46336364746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15370, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.2996681421995163, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.12033190205693245, + "grad_norm": 0.00030349052394740283, + "kept_token_count": 572.975, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886419296264648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4367638838593845e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5136413574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.69553833007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15380, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2995427668094635, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.1255710981786251, + "grad_norm": 0.0003059523005504161, + "kept_token_count": 573.1515625, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884539365768432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.510198887146544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8657928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.28420104980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15390, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3000594645738602, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.1300327055156231, + "grad_norm": 0.00018843649013433605, + "kept_token_count": 572.9578125, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888577044010163, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.352478717919439e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.01947021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8163635253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15400, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3008943647146225, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.12214994207024574, + "grad_norm": 0.00023503311967942864, + "kept_token_count": 572.021875, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885750055313111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.462905744730961e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2778839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.67046813964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15410, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29968231320381167, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.1429439403116703, + "grad_norm": 0.00025444646598771214, + "kept_token_count": 573.14375, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889472305774689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.317505226936191e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.16407165527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2196380615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15420, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.30001414120197295, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.12199889346957207, + "grad_norm": 0.00027842921554110944, + "kept_token_count": 572.846875, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882843971252442, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.576424216793384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6427856445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5679229736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15430, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.3000706732273102, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.12978129237890243, + "grad_norm": 0.00024165913055185229, + "kept_token_count": 572.859375, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884167850017548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.524715113802813e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4379852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7067077636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15440, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.29879834651947024, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.12345970645546914, + "grad_norm": 0.00021589064272120595, + "kept_token_count": 573.871875, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0, + "mgd_cos_sim": 0.98806112408638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.663638574129436e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3062255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.94135131835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15450, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3001187592744827, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.13145839869976045, + "grad_norm": 0.0002713932190090418, + "kept_token_count": 572.58125, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885450184345246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4746156709152275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4599945068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8427276611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15460, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2996497184038162, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.12193767949938775, + "grad_norm": 0.0002263287897221744, + "kept_token_count": 573.134375, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887546420097351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.39273688243702e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.83094177246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.96047058105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30011466443538665, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.13547885566949844, + "grad_norm": 0.0001941250666277483, + "kept_token_count": 572.3015625, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988745641708374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3962504059891215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.8017852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2738922119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15480, + "valid_token_count": 817.709375 + }, + { + "actual_token_mask_ratio": 0.30115934312343595, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.1264052964746952, + "grad_norm": 0.00032796175219118595, + "kept_token_count": 571.74375, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885914564132691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.456472233869135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.90704956054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9778533935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15490, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2991918295621872, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.12302099242806434, + "grad_norm": 0.00024043199664447457, + "kept_token_count": 573.3453125, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988336181640625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.556197091005742e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.98448486328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9249572753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15500, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.2999932706356049, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.12695189490914344, + "grad_norm": 0.0002611284435261041, + "kept_token_count": 572.959375, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887295842170716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4025299575878305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9222869873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9923522949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15510, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.30109443962574006, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.13606305047869682, + "grad_norm": 0.0004303155583329499, + "kept_token_count": 571.7703125, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887596249580384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.390788344608154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2522308349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.86163330078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15520, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3004150062799454, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.12777236402034758, + "grad_norm": 0.0003105327778030187, + "kept_token_count": 572.353125, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896007359027863, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.062233019794803e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.21715393066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9969757080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15530, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.300687313079834, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.1278661794960499, + "grad_norm": 0.00036166032077744603, + "kept_token_count": 571.921875, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988041615486145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.671262613555882e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9541076660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.213671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15540, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.30008154213428495, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.13152410686016083, + "grad_norm": 0.0001801577309379354, + "kept_token_count": 572.9640625, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889274358749389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.325245245127007e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.350927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.47157592773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15550, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.301356366276741, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.13751614689826966, + "grad_norm": 0.0003112312697339803, + "kept_token_count": 571.54375, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888348400592804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3614090827759355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1280944824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2495971679688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15560, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30040723979473116, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.13139218464493752, + "grad_norm": 0.00024904974270612, + "kept_token_count": 572.4765625, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882874608039856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.575228522298858e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.41844177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1847198486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15570, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30005563199520113, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.1309104658663273, + "grad_norm": 0.0002299516199855134, + "kept_token_count": 573.1234375, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889417946338653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.319630897953175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9922729492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20241088867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15580, + "valid_token_count": 818.815625 + }, + { + "actual_token_mask_ratio": 0.29913117587566374, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.11475575864315032, + "grad_norm": 0.00020265972125343978, + "kept_token_count": 573.55625, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886533260345459, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4323131442070006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.61053771972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9423126220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15590, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.30002419650554657, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.1282272644340992, + "grad_norm": 0.000265238166321069, + "kept_token_count": 573.015625, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881932735443115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.61202365841018e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7795166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9474334716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15600, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.301050078868866, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.13592597991228103, + "grad_norm": 0.00028272089548408985, + "kept_token_count": 572.2078125, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888623833656311, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.350649724074174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4763427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.42719116210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15610, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.2994924306869507, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.12517881467938424, + "grad_norm": 0.00022589703439734876, + "kept_token_count": 573.2265625, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988575679063797, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.462645483727101e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1961242675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4484069824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15620, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.3007481038570404, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.12326701804995537, + "grad_norm": 0.00022983764938544482, + "kept_token_count": 572.209375, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887466073036194, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3958762762486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9534484863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15630, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2999395102262497, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.13012229427695274, + "grad_norm": 0.00029315645224414766, + "kept_token_count": 573.0078125, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881254374980927, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.638519931177143e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.02500915527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3354034423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15640, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.3009347438812256, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.12376493290066719, + "grad_norm": 0.00025252593331970274, + "kept_token_count": 572.0203125, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988779878616333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3828782145283186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3448120117188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2887817382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15650, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.3002598524093628, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.13035803511738778, + "grad_norm": 0.00020891152962576598, + "kept_token_count": 572.7578125, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886215150356292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.444739061000291e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9754577636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.66294555664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15660, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.30033194422721865, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.12383319661021233, + "grad_norm": 0.0002824626863002777, + "kept_token_count": 572.7, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885419607162476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.475815949263051e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1099365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7114654541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15670, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.29996931552886963, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.1294195257127285, + "grad_norm": 0.0002431937027722597, + "kept_token_count": 572.778125, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888268232345581, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.364536471257452e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7830749511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8031341552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.299161759018898, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.12946044355630876, + "grad_norm": 0.00022691153571940958, + "kept_token_count": 573.446875, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988889193534851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3401808579801585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.58063049316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.52858581542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15690, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3001391112804413, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.13402158692479132, + "grad_norm": 0.00023998577671591192, + "kept_token_count": 572.5578125, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875557959079743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.8610366866341793e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4406433105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9968688964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15700, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.3000844955444336, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.12453240528702736, + "grad_norm": 0.0002490276237949729, + "kept_token_count": 572.5140625, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887941122055054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.377323421067558e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.80507507324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9333465576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15710, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.30059874355793, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.12419676184654235, + "grad_norm": 0.0003740007523447275, + "kept_token_count": 572.35625, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888407945632934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.359082086011767e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.555908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2707916259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30041842460632323, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.126888919621706, + "grad_norm": 0.0002817627973854542, + "kept_token_count": 572.490625, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884605765342712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5076034439262e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.9818450927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.83392639160155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15730, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3008354067802429, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.12283688113093376, + "grad_norm": 0.0003128347161691636, + "kept_token_count": 571.9875, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884334444999695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.518200948950834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.13380126953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.15931091308596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15740, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30031055510044097, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.126824302226305, + "grad_norm": 0.00019882961350958794, + "kept_token_count": 572.3046875, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893033862113952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.178386188868899e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.55507202148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.379443359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15750, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.30056596398353574, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.12595481500029565, + "grad_norm": 0.0003430317447055131, + "kept_token_count": 572.2828125, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9879418194293976, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.710250577772967e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.81324768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.58798828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15760, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30019655227661135, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.13532312959432602, + "grad_norm": 0.0002731103450059891, + "kept_token_count": 572.5734375, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889794290065765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.304931389924605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2639556884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3454162597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.29891806542873384, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.12948136553168296, + "grad_norm": 0.00022435191203840077, + "kept_token_count": 573.4609375, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889226853847504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.32709523010999e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.05210571289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89117126464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15780, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.29987166821956635, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.13215796276926994, + "grad_norm": 0.00020314303401391953, + "kept_token_count": 572.8796875, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891613841056823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.233856998325791e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2714324951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5456878662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3004288852214813, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.12079053074121475, + "grad_norm": 0.00023472492466680706, + "kept_token_count": 572.6453125, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0, + "mgd_cos_sim": 0.98872691988945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.403564780659508e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.176953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.63963623046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15800, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.30020529627799986, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.1303139604628086, + "grad_norm": 0.00023249330115504563, + "kept_token_count": 572.4859375, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885586738586426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4692812298308124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.1485931396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06976318359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15810, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30060163140296936, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.1267813414335251, + "grad_norm": 0.0002519371046219021, + "kept_token_count": 572.203125, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884799361228943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.500047252804506e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.30348510742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3230682373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15820, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29936585426330564, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.12074542492628097, + "grad_norm": 0.00020688406948465854, + "kept_token_count": 573.315625, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893015682697296, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.179094321443699e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5345062255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.98659057617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15830, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30059259235858915, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.12782954126596452, + "grad_norm": 0.00034734528162516654, + "kept_token_count": 572.315625, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9874760329723358, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.892194774583913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7269348144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9982849121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15840, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30017063617706297, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.13251445069909096, + "grad_norm": 0.00024131726240739226, + "kept_token_count": 572.60625, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891534090042114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.236968161421828e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7810577392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5531280517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15850, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2999095290899277, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.1290126547217369, + "grad_norm": 0.0002615912235341966, + "kept_token_count": 572.903125, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886706888675689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.425531878951006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.82489318847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.89504089355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15860, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2996039241552353, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.12822115421295166, + "grad_norm": 0.0003029083018191159, + "kept_token_count": 573.0453125, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891554415225983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.236171844240744e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3682800292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.05120544433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15870, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.299346336722374, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.12812134698033334, + "grad_norm": 0.0002929873298853636, + "kept_token_count": 573.3359375, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887181580066681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.406989028211683e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.01537170410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2433319091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15880, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2996413350105286, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.12905998453497886, + "grad_norm": 0.00019926724780816585, + "kept_token_count": 572.821875, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888871788978577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.34096589742694e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9400604248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0330780029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15890, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.3003657698631287, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.1284607969224453, + "grad_norm": 0.00018632065621204674, + "kept_token_count": 572.3765625, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886697709560395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.425881379575003e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.96327209472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72779235839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15900, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.3007241189479828, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.12983943819999694, + "grad_norm": 0.0002737348258960992, + "kept_token_count": 572.4171875, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890018939971924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.296153856557794e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.50152587890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2078552246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15910, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.29964481890201566, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.1357685148715973, + "grad_norm": 0.0002022729313466698, + "kept_token_count": 573.2765625, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881805360317231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.617000304278918e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1720001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15920, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.30021971464157104, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.12732096761465073, + "grad_norm": 0.00034970982233062387, + "kept_token_count": 572.7703125, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9875002443790436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.882730536337476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3284088134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9729461669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15930, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.3003643125295639, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.12476528435945511, + "grad_norm": 0.0003155261219944805, + "kept_token_count": 572.6484375, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892667174339295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.192714586679358e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1435516357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9505950927734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15940, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.30017839968204496, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.13163234367966653, + "grad_norm": 0.0003421029541641474, + "kept_token_count": 572.7625, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989200234413147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2186720747849904e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6145416259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3925811767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15950, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30026220679283144, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.12704163938760757, + "grad_norm": 0.00022249265748541802, + "kept_token_count": 572.7203125, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884072363376617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.528433819359634e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4666381835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3004669189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15960, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2995722949504852, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.13129559829831122, + "grad_norm": 0.00028844166081398726, + "kept_token_count": 573.3203125, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895030558109283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1003943988471295e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9507568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.026904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15970, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30022935271263124, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.13591290786862373, + "grad_norm": 0.00022672938939649612, + "kept_token_count": 572.596875, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887585759162902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.391197035147343e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.06987915039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9985870361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15980, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29974051415920255, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.1277748316526413, + "grad_norm": 0.00022774194076191634, + "kept_token_count": 573.10625, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880677163600922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6610689605586233e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1034271240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2112213134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15990, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.30103454887866976, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.1248002141714096, + "grad_norm": 0.00024478783598169684, + "kept_token_count": 572.1625, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891015112400054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.257236942066811e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.99521789550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7417755126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16000, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.30046021938323975, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.12030631378293037, + "grad_norm": 0.0002277879830216989, + "kept_token_count": 572.2484375, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988284456729889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.576405008265283e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8871704101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8468902587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16010, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.2991674095392227, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.13482441529631614, + "grad_norm": 0.00023609667550772429, + "kept_token_count": 573.325, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883555054664612, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.54864861239912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2894653320312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.83290100097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16020, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.2992387503385544, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.12054355591535568, + "grad_norm": 0.00023710392997600138, + "kept_token_count": 573.5765625, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889994442462922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2971154471160844e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4538543701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6015350341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16030, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.2987727165222168, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.13323130533099176, + "grad_norm": 0.00023799201881047338, + "kept_token_count": 573.8484375, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890301525592804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.28511430072831e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.477197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79710693359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16040, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.29969831705093386, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.13017982840538025, + "grad_norm": 0.00019616477948147804, + "kept_token_count": 572.9453125, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887921810150146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.378072808322031e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3510986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9455291748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2999586403369904, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.1285386361181736, + "grad_norm": 0.00039330386789515615, + "kept_token_count": 572.603125, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892352819442749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.204981160000898e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3538757324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.28278198242185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16060, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.29998726546764376, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.13250125348567962, + "grad_norm": 0.00025434992858208716, + "kept_token_count": 572.90625, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887543499469758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.392848422867246e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1677276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3136413574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.30014287531375883, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.12549526765942573, + "grad_norm": 0.00028371316147968173, + "kept_token_count": 572.5203125, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889589965343475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3129095865879206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.32326049804686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.63463439941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16080, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.2997744709253311, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.13382804319262503, + "grad_norm": 0.0002417774376226589, + "kept_token_count": 572.7609375, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889517784118652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.315726000641007e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.90447387695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5291473388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16090, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.29871380627155303, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.11692189797759056, + "grad_norm": 0.00021045078756287694, + "kept_token_count": 573.83125, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988204973936081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.60745479358593e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.15728149414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.59153137207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16100, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29967832565307617, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.12955221235752107, + "grad_norm": 0.00019726371101569384, + "kept_token_count": 572.9046875, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988779479265213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3830326831084676e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7047424316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.92344970703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16110, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.2996951311826706, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.12507453188300133, + "grad_norm": 0.0002175591216655448, + "kept_token_count": 573.0609375, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887456595897675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3962504059891215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.80567932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55248107910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16120, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3002811670303345, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.12525676041841508, + "grad_norm": 0.00035972989280708134, + "kept_token_count": 572.4890625, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883066117763519, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.567745309032034e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6126281738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.92896728515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16130, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30039840638637544, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.1300415076315403, + "grad_norm": 0.0002854359336197376, + "kept_token_count": 572.409375, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885914027690887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.456496244529262e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9816955566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6306610107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16140, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2999484747648239, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.13017498925328255, + "grad_norm": 0.0002361617371207103, + "kept_token_count": 572.46875, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888561010360718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3531023766263385e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6265045166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2818359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16150, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.30005061328411103, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.12838463634252548, + "grad_norm": 0.00026165044982917607, + "kept_token_count": 572.51875, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891221046447753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.249196354066953e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.15346069335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.55567932128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16160, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.29912366569042204, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.12284273654222488, + "grad_norm": 0.00024444039445370436, + "kept_token_count": 573.6359375, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989140921831131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.241847200319171e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9293640136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1330139160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29911835491657257, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.13344804123044013, + "grad_norm": 0.00031489538378082216, + "kept_token_count": 573.5578125, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881891727447509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.613619130395818e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8173492431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.40733032226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16180, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2990534156560898, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.1324528709053993, + "grad_norm": 0.00026248645735904574, + "kept_token_count": 573.5671875, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893717408180237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.151676839683205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7802490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0788238525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16190, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3002701371908188, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.1319503016769886, + "grad_norm": 0.0002654271374922246, + "kept_token_count": 572.4609375, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881748616695404, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.619214378180914e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5163177490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.90366516113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16200, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2993286609649658, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.13369750529527663, + "grad_norm": 0.0002243041672045365, + "kept_token_count": 573.3421875, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9880183935165405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.680335441662464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1522491455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.73622741699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16210, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2986005783081055, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.13285831362009048, + "grad_norm": 0.00024438765831291676, + "kept_token_count": 573.878125, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989306902885437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.177016198809724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.97492065429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.72820434570315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16220, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3001225501298904, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.12314685359597206, + "grad_norm": 0.00018676753097679466, + "kept_token_count": 572.6453125, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893983364105224, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.141292920394335e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1528381347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6299530029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3000183284282684, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.12476552650332451, + "grad_norm": 0.00023275606508832425, + "kept_token_count": 572.96875, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892673969268799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1924430115614085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41526184082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.32733154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16240, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.2987651199102402, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.12996652573347092, + "grad_norm": 0.0001973547914531082, + "kept_token_count": 573.903125, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989386010169983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.146110695728567e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4389984130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2076354980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16250, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.3003535747528076, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.12883174493908883, + "grad_norm": 0.00021917330741416663, + "kept_token_count": 572.3734375, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885802149772644, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.460870877665002e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0542236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2555633544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16260, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30004623234272004, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.13327017799019814, + "grad_norm": 0.00021741313685197383, + "kept_token_count": 572.9078125, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891085505485535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2544870666461065e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3374267578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4677673339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16270, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.3003967136144638, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.1291326865553856, + "grad_norm": 0.00018460306455381215, + "kept_token_count": 572.571875, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886826157569886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.420875484356657e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7808532714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3882995605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16280, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.3003389030694962, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.13544081896543503, + "grad_norm": 0.00020443648099899292, + "kept_token_count": 572.4765625, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888508021831512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.355175406089984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8944030761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.21018676757814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2996924608945847, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.12711079865694047, + "grad_norm": 0.00019960560894105583, + "kept_token_count": 573.284375, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890855431556702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2634837154764685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2798767089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.490576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16300, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.30075606405735017, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.11625799536705017, + "grad_norm": 0.00020087572920601815, + "kept_token_count": 572.0109375, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890709519386292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.269177261448931e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7666931152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9200103759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16310, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.29982773065567014, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.13623644337058066, + "grad_norm": 0.0002558884443715215, + "kept_token_count": 573.0390625, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886802554130554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.421793310029898e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5697906494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.68046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16320, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.3002051055431366, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.12336885929107666, + "grad_norm": 0.0002699170436244458, + "kept_token_count": 572.6796875, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892877876758576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.184479075775016e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.36399536132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.36255798339846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16330, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3001981258392334, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.1386720821261406, + "grad_norm": 0.0002137223636964336, + "kept_token_count": 572.528125, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988446980714798, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.512913765211124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.46022338867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9458312988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16340, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2993430644273758, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.12512403428554536, + "grad_norm": 0.00021568052761722356, + "kept_token_count": 573.2953125, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891147673130035, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2520591523498294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2706024169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.71298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16350, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.30052924156188965, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.136213269084692, + "grad_norm": 0.0003551016852725297, + "kept_token_count": 572.284375, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885172724723816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.485458011913579e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8746063232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4626953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16360, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.29971829056739807, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.12412977293133735, + "grad_norm": 0.0002270096738357097, + "kept_token_count": 573.0125, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892772376537323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.18859752244316e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7342590332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65887756347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16370, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.299522140622139, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.13308271691203116, + "grad_norm": 0.00020884955301880836, + "kept_token_count": 573.1375, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889061570167541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3335566806490534e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8393157958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3761474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16380, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3003991305828094, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.1321195788681507, + "grad_norm": 0.0002471821499057114, + "kept_token_count": 572.43125, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885524928569793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4716992488247345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0924377441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.20021667480466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16390, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30062372982501984, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.1331729620695114, + "grad_norm": 0.00019970789435319602, + "kept_token_count": 572.1984375, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891630172729492, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2332163502578624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.470361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9245666503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16400, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.3002725601196289, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.12756994739174843, + "grad_norm": 0.0002648691588547081, + "kept_token_count": 572.590625, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893247842788696, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.170022984908428e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.62828674316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8812652587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16410, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.3003775328397751, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.1242927648127079, + "grad_norm": 0.0001997230137931183, + "kept_token_count": 572.4859375, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888684034347535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.348303737060633e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5144744873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0299346923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.3006232470273972, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.12017521187663079, + "grad_norm": 0.00019087950931861997, + "kept_token_count": 572.440625, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892271637916565, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.208164718875196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9078063964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.00037841796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16430, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.29971493780612946, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.12538672834634781, + "grad_norm": 0.00017268565716221929, + "kept_token_count": 573.121875, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9901015043258667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.866618135361932e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2475280761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.41681213378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16440, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29980514049530027, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.12930695489048957, + "grad_norm": 0.0002526336465962231, + "kept_token_count": 572.8953125, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883588433265686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5473424324882214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3849212646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1434661865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16450, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3001714915037155, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.11904852315783501, + "grad_norm": 0.00020525188301689923, + "kept_token_count": 572.846875, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893512308597565, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.159692834946327e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6277221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5448913574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16460, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.2998546302318573, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.13190807178616523, + "grad_norm": 0.0002089549379888922, + "kept_token_count": 573.0859375, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890890181064605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2621183092705904e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.99518127441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.95103454589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16470, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.2994737684726715, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.13310555070638658, + "grad_norm": 0.00026133106439374387, + "kept_token_count": 573.19375, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885467052459717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.473959234019276e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.67470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.44956665039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16480, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2989550203084946, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.127498160302639, + "grad_norm": 0.00021990340610500425, + "kept_token_count": 573.7359375, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893286466598511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.168517807556782e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2734649658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1976684570312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16490, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3002693474292755, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.12241698503494262, + "grad_norm": 0.00020315377332735807, + "kept_token_count": 572.684375, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894550740718842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.119126679142937e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.67943115234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0882965087891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16500, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29927873611450195, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.1313516341149807, + "grad_norm": 0.00021169842511881143, + "kept_token_count": 573.5078125, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883490324020385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.551171768980566e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9954132080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1649658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16510, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2998234897851944, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.11734630092978478, + "grad_norm": 0.00021567671501543373, + "kept_token_count": 572.85625, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895754992961884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.072090050613042e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.78455810546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3132049560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16520, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3003904312849045, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.1286807306110859, + "grad_norm": 0.00035811957786791027, + "kept_token_count": 572.51875, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892623603343964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1944133045035414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.98407592773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3105865478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16530, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2998813450336456, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.12306973934173585, + "grad_norm": 0.00020981575653422624, + "kept_token_count": 572.909375, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884692966938019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.504196404013783e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.37816467285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.10853576660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.29972773492336274, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.12958128228783608, + "grad_norm": 0.0002412401227047667, + "kept_token_count": 573.153125, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889519393444062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.315668265917338e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8994537353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5838989257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16550, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30017792880535127, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.12436853125691413, + "grad_norm": 0.00025528506375849247, + "kept_token_count": 572.675, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892631947994233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1940863593481484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2735198974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.69627380371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2993337333202362, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.12583415657281877, + "grad_norm": 0.0002237942098872736, + "kept_token_count": 573.465625, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884327411651611, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.51848089142004e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3109558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1763031005859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16570, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2994515597820282, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.1303768925368786, + "grad_norm": 0.00028432797989808023, + "kept_token_count": 573.765625, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890487551689148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2778513306984675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.35594177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.79090881347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16580, + "valid_token_count": 819.025 + }, + { + "actual_token_mask_ratio": 0.2997803151607513, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.12741099894046784, + "grad_norm": 0.0002468363381922245, + "kept_token_count": 572.8015625, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892835736274719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.186128680885304e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.81763916015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.89283447265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16590, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29966571033000944, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.1315435953438282, + "grad_norm": 0.00021862381254322827, + "kept_token_count": 573.3, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876732170581818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.815171305381227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.5261291503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0278259277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16600, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.29876281023025514, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.12361611053347588, + "grad_norm": 0.0003534049028530717, + "kept_token_count": 573.840625, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890678465366364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2703942744992675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8188781738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.83245544433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16610, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.30062511265277864, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.131841978430748, + "grad_norm": 0.00021259723871480674, + "kept_token_count": 572.16875, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884262084960938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5210296229925005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.83421020507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0110229492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16620, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3000632345676422, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.12835498824715613, + "grad_norm": 0.00023058919759932905, + "kept_token_count": 572.8578125, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893122375011444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1749259617063216e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2479675292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3482269287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16630, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.3004313826560974, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.13667130172252656, + "grad_norm": 0.00022598613577429205, + "kept_token_count": 572.615625, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886225879192352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4443250953918326e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.893212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2856750488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16640, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29927281439304354, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.13117756098508834, + "grad_norm": 0.00021524603653233498, + "kept_token_count": 573.33125, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886687457561493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.426294508448336e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1050598144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.75816345214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16650, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2984954953193665, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.13070956170558928, + "grad_norm": 0.0002867162402253598, + "kept_token_count": 574.1140625, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989111602306366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.253297847753856e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.36670837402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6447387695313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16660, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29992190897464754, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.13703641593456267, + "grad_norm": 0.00023706232605036348, + "kept_token_count": 572.8984375, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887273848056793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4033903395757076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.94642944335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.62640075683595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3007345050573349, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.12383622527122498, + "grad_norm": 0.00021893170196563005, + "kept_token_count": 572.384375, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894236862659455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1313909241580406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.12245788574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4395355224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16680, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2996480196714401, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.13175617381930352, + "grad_norm": 0.00023794635490048677, + "kept_token_count": 573.178125, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890166878700256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.290384604246356e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.27330627441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8745635986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16690, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2994959205389023, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.12957584708929062, + "grad_norm": 0.0002516099775675684, + "kept_token_count": 573.3609375, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889725744724274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.307610433897935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2962371826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.18284606933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16700, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.2991768687963486, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.12898639366030692, + "grad_norm": 0.0002466146252118051, + "kept_token_count": 573.5515625, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889152765274047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3299847675370984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8165710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.63295288085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16710, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2997430384159088, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.12251501753926278, + "grad_norm": 0.0002071356138912961, + "kept_token_count": 573.0234375, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891008853912353, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.257483051333111e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.16437072753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8070983886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16720, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2999322384595871, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.12903321757912636, + "grad_norm": 0.000336356257321313, + "kept_token_count": 572.834375, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883270263671875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.559773660730571e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4303009033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.612744140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16730, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30062898695468904, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.12292448058724403, + "grad_norm": 0.0002113314694724977, + "kept_token_count": 572.5109375, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897560894489288, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.001547677034978e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.25731201171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.15440368652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16740, + "valid_token_count": 818.609375 + }, + { + "actual_token_mask_ratio": 0.3002153515815735, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.11983055621385574, + "grad_norm": 0.00021345990535337478, + "kept_token_count": 572.6875, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896290957927704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.051153591717593e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6615783691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.161962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16750, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2998014479875565, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.1382945731282234, + "grad_norm": 0.00027506539481692016, + "kept_token_count": 573.240625, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891056954860687, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2556084736133926e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.14024658203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61253662109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16760, + "valid_token_count": 818.6890625 + }, + { + "actual_token_mask_ratio": 0.3009143054485321, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.12698885276913643, + "grad_norm": 0.00024588240194134414, + "kept_token_count": 572.3765625, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989428848028183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.129370681766886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.62732849121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.72275390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16770, + "valid_token_count": 818.7546875 + }, + { + "actual_token_mask_ratio": 0.30043852925300596, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.1282515950500965, + "grad_norm": 0.00019764482567552477, + "kept_token_count": 572.5640625, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896436035633087, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.045487512485124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.43711242675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7542358398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16780, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.30081731379032134, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.12006961703300476, + "grad_norm": 0.00021025491878390312, + "kept_token_count": 572.040625, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989815229177475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9784450564184226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8311767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6580810546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16790, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3000706732273102, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.1248830646276474, + "grad_norm": 0.00031497798045165837, + "kept_token_count": 572.8328125, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893301129341125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.167948827671353e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1087677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.53368225097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16800, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2996141850948334, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.12656928971409798, + "grad_norm": 0.0002112884831149131, + "kept_token_count": 573.090625, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989975494146347, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.915838860848453e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7625427246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4199615478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.29871382415294645, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.12326544895768166, + "grad_norm": 0.0003018562856595963, + "kept_token_count": 573.7828125, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894799649715423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1094116750173273e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.800244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0566040039063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16820, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.30052360594272615, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.13211213275790215, + "grad_norm": 0.00026288829394616187, + "kept_token_count": 572.3171875, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889589846134186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.312915989430621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.53502197265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5031494140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16830, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2998712569475174, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.13359716162085533, + "grad_norm": 0.00021862430730834603, + "kept_token_count": 573.2015625, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894018113613129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.139935554121621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.14684143066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.04393615722654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16840, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.3008388549089432, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.12869543358683586, + "grad_norm": 0.00024633988505229354, + "kept_token_count": 572.1921875, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888590216636658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3519659811863676e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.59374084472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0644897460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.300018310546875, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.1204471841454506, + "grad_norm": 0.00030819824314676225, + "kept_token_count": 572.734375, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882895588874817, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5744051385554484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.263525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.73646850585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16860, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.299224254488945, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.12410595566034317, + "grad_norm": 0.00029125530272722244, + "kept_token_count": 573.4734375, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887261807918548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4038527630618776e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8098999023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.75169067382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16870, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30041050910949707, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.12290723323822021, + "grad_norm": 0.0002984503225889057, + "kept_token_count": 572.6234375, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884183347225189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5241063344292345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.54437561035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3383483886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16880, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2981529086828232, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.11889078840613365, + "grad_norm": 0.0002270438417326659, + "kept_token_count": 574.4546875, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989760822057724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.999701402790379e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.85080871582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7691131591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16890, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.29940691590309143, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.1292151466012001, + "grad_norm": 0.00017509715689811856, + "kept_token_count": 573.096875, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889386832714081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.320847874623723e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.57928161621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0816955566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16900, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.2995512366294861, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.1212182179093361, + "grad_norm": 0.0002682124904822558, + "kept_token_count": 573.490625, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898008942604065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.98404969018884e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.19285888671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.14788208007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16910, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.2994286924600601, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.130429707467556, + "grad_norm": 0.00024322676472365856, + "kept_token_count": 573.2953125, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989467179775238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.114403636776842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0531921386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7776580810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16920, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.29969464242458344, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.1235724076628685, + "grad_norm": 0.0003182432265020907, + "kept_token_count": 573.125, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988336569070816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.556040112220216e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0005187988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2066345214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16930, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.29955693483352663, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.12267159670591354, + "grad_norm": 0.00020602144650183618, + "kept_token_count": 573.209375, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887136697769165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.408734785101842e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.38079223632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.03785400390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16940, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.299658277630806, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.12817564904689788, + "grad_norm": 0.0002912531781475991, + "kept_token_count": 573.1046875, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888450145721436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.357438810984604e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.48207092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.90904846191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16950, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2999594658613205, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.11586862653493882, + "grad_norm": 0.00020148260227870196, + "kept_token_count": 572.821875, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891714334487915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.229928526910953e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7250061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.6878997802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16960, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2989752471446991, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.1284421995282173, + "grad_norm": 0.0001924349635373801, + "kept_token_count": 573.7796875, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883555591106414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.548626275209244e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.68479614257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3354156494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16970, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.3004461914300919, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.13573400527238846, + "grad_norm": 0.00019308968330733478, + "kept_token_count": 572.2953125, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885287582874298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.480972420424223e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41737670898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0290466308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16980, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.29977835416793824, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.13257328644394875, + "grad_norm": 0.0002167311467928812, + "kept_token_count": 573.0234375, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891656160354614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.23219953518128e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9215484619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0830902099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16990, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2998905718326569, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.13108196407556533, + "grad_norm": 0.0002039739047177136, + "kept_token_count": 572.7828125, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892730057239533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1902519660652616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.59352416992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.99492797851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17000, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2997723609209061, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.12993529289960862, + "grad_norm": 0.000213274935958907, + "kept_token_count": 572.865625, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886744618415833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.424054459377658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.38206176757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8035522460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17010, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30011485815048217, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.12392206341028214, + "grad_norm": 0.00023319780302699655, + "kept_token_count": 572.7453125, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9883776545524597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.539998008112889e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8422088623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1029022216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17020, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30047589242458345, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.12625968009233474, + "grad_norm": 0.00018476274271961302, + "kept_token_count": 572.315625, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896909534931183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.026987371616997e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.22756042480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1154479980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17030, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2994010835886002, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.13029197752475738, + "grad_norm": 0.0004204704600851983, + "kept_token_count": 573.1921875, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884389638900757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.516051521932241e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.40196533203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3682556152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17040, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2991786003112793, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.1262093171477318, + "grad_norm": 0.0003499820304568857, + "kept_token_count": 573.40625, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889144957065582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.330304109316785e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.52826232910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.96705322265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17050, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.3002432346343994, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.12152314260601997, + "grad_norm": 0.00025408650981262326, + "kept_token_count": 572.459375, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889372408390045, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.321405904192943e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3949676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3767150878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17060, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3003293514251709, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.1232435680925846, + "grad_norm": 0.00023405709362123162, + "kept_token_count": 572.4796875, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881057381629944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.646215384127572e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9150695800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4335510253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17070, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2999785304069519, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.12625957429409027, + "grad_norm": 0.00024511502124369144, + "kept_token_count": 572.8671875, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895028114318848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.100486439710949e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8914733886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7758331298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30070590376853945, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.12901946306228637, + "grad_norm": 0.000271544442512095, + "kept_token_count": 572.1109375, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989213615655899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.21345259383088e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5914703369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.76414794921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17090, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.2996858835220337, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.11429238542914391, + "grad_norm": 0.0002644642081577331, + "kept_token_count": 572.8640625, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892739593982697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.189884311927017e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.41778869628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.43487854003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.2994845241308212, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.13991707563400269, + "grad_norm": 0.00026135853840969503, + "kept_token_count": 573.1515625, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889518082141876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.315717924328055e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4762268066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.29017639160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17110, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2987507045269012, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.11643002852797509, + "grad_norm": 0.00021627178648486733, + "kept_token_count": 573.9171875, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892625927925109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.194314496999141e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5958221435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4517120361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17120, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2999760568141937, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.1328433185815811, + "grad_norm": 0.0002250561665277928, + "kept_token_count": 572.896875, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889792382717133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3049991290899925e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.85346069335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6121856689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17130, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30029223263263705, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.12290805503726006, + "grad_norm": 0.00030192098347470164, + "kept_token_count": 572.2671875, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895268619060517, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.091088776476681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.15602416992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.21917114257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17140, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.29987863898277284, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.12513653188943863, + "grad_norm": 0.00020312560081947595, + "kept_token_count": 572.646875, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893024682998657, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.178746858087834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.63341369628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.12919921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17150, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.29957888424396517, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.12104695662856102, + "grad_norm": 0.00019994248577859253, + "kept_token_count": 573.2234375, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898974180221558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.946341021219268e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2594879150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0983489990234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17160, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.30058057606220245, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.13302091807126998, + "grad_norm": 0.00021441091666929424, + "kept_token_count": 572.20625, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892974555492401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.18069746956462e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9578094482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0711639404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17170, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.299591001868248, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.1365131124854088, + "grad_norm": 0.0003046166675630957, + "kept_token_count": 573.1875, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897243082523346, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0139638440450656e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2747009277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.849658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17180, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.30097249448299407, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.12301400303840637, + "grad_norm": 0.0002161234588129446, + "kept_token_count": 571.94375, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897212266921998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.015163794974796e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9731079101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4943389892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17190, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29879132509231565, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.1377164125442505, + "grad_norm": 0.0002739510382525623, + "kept_token_count": 573.6, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988772201538086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.385874017316382e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41651306152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.51987609863284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17200, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.30049148499965667, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.1261010594666004, + "grad_norm": 0.00021252375154290348, + "kept_token_count": 572.5015625, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888685584068299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.348242837295402e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.70928344726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1138458251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17210, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2999220937490463, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.14165103435516357, + "grad_norm": 0.0003159102634526789, + "kept_token_count": 572.7140625, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890113890171051, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2924440276692624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4209869384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17220, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.29945302903652193, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.1404708169400692, + "grad_norm": 0.00022806103515904397, + "kept_token_count": 573.165625, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889550089836121, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3144723531440834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.29918518066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.14666137695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17230, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2994943499565125, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.132638331502676, + "grad_norm": 0.0002930336631834507, + "kept_token_count": 573.521875, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896537959575653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.041512838739436e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.81264038085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9711059570312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17240, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.29931686222553255, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.12477863878011704, + "grad_norm": 0.00021813716739416122, + "kept_token_count": 573.3765625, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895446062088012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.084155079908669e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8581604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1529083251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17250, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2986567854881287, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.13368840217590333, + "grad_norm": 0.0002870313182938844, + "kept_token_count": 573.71875, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892074823379516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2158506039413625e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7515380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7602569580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17260, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.3002653121948242, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.1253958061337471, + "grad_norm": 0.0002444951969664544, + "kept_token_count": 572.75625, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885315537452698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4798814633395524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.881005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0375061035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17270, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.29916794300079347, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.12383925467729569, + "grad_norm": 0.00021145773644093424, + "kept_token_count": 573.371875, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9903495013713837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.769740869756788e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7706665039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9479248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.300563833117485, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.12643380463123322, + "grad_norm": 0.00037283854908309877, + "kept_token_count": 571.9421875, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888294279575348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3635256952256894e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.23980407714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2274566650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17290, + "valid_token_count": 817.721875 + }, + { + "actual_token_mask_ratio": 0.30084024369716644, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.12096915245056153, + "grad_norm": 0.0001741753367241472, + "kept_token_count": 572.2359375, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898076355457306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.98141513869632e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5178466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8768646240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17300, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.29979884028434756, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.12674434185028077, + "grad_norm": 0.00023851484002079815, + "kept_token_count": 573.0578125, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0, + "mgd_cos_sim": 0.98843132853508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.519029789662454e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6108367919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9474182128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17310, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2996334761381149, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.13244456127285958, + "grad_norm": 0.00023013564350549132, + "kept_token_count": 572.859375, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892804026603699, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.187361046206206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3576385498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0686798095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17320, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.30100518465042114, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.1259379953145981, + "grad_norm": 0.00018389476463198662, + "kept_token_count": 572.0390625, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891295492649078, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2462904821150006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6518524169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3662445068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17330, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30122811794281007, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.12949502021074294, + "grad_norm": 0.0002468627935741097, + "kept_token_count": 571.7328125, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887104868888855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.409982648212463e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1708465576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.33716735839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17340, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2990674674510956, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.12675339207053185, + "grad_norm": 0.00023258384317159653, + "kept_token_count": 573.5125, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895157516002655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.095425028936006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.311669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9927917480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3003229945898056, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.11835083067417144, + "grad_norm": 0.0003445657785050571, + "kept_token_count": 572.6265625, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895249426364898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.091838673048187e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.23817443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6607421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17360, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.29925046265125277, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.1356496073305607, + "grad_norm": 0.00024111499078571796, + "kept_token_count": 573.5234375, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893395900726318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1642368887551126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8613677978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34119873046876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17370, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3002016872167587, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.12634653151035308, + "grad_norm": 0.0004143999540247023, + "kept_token_count": 572.7890625, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889389753341675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.320732368796598e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.09007873535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.58941345214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17380, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.3003446400165558, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.12933584153652192, + "grad_norm": 0.00023043269175104797, + "kept_token_count": 572.2890625, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989332765340805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.166912185610272e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.58665771484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.56434326171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17390, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.3008113205432892, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.13323336094617844, + "grad_norm": 0.0002123690937878564, + "kept_token_count": 571.8640625, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989294558763504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.181834483460989e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7600341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.58995666503904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17400, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.29918268620967864, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.1267825372517109, + "grad_norm": 0.0002810238511301577, + "kept_token_count": 573.4390625, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891070067882538, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.255093408573884e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6985687255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75269775390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3002029538154602, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.12772505804896356, + "grad_norm": 0.00030951944063417614, + "kept_token_count": 572.4296875, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9886296153068542, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4415756565285844e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9453369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6892791748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17420, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.3005447953939438, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.1286654733121395, + "grad_norm": 0.0002474238572176546, + "kept_token_count": 572.5203125, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894985496997833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.102149250684306e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3352569580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.99830627441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17430, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.30030189752578734, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.1289424829185009, + "grad_norm": 0.00024808302987366915, + "kept_token_count": 572.6359375, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892986178398132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.180242831353098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7621154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0942108154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17440, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.299425208568573, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.11846935823559761, + "grad_norm": 0.00026125198928639293, + "kept_token_count": 573.178125, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9903415501117706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.772851887333673e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.17771911621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7781158447266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17450, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2994793504476547, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.12963199317455293, + "grad_norm": 0.00023494349443353713, + "kept_token_count": 573.0828125, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894664645195007, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1146816511172804e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5203521728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8162445068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17460, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30050547420978546, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.1267831139266491, + "grad_norm": 0.0003398352419026196, + "kept_token_count": 572.5109375, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9885616958141327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.468103143153712e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.65403747558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.59136962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17470, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.3000269681215286, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.13043006658554077, + "grad_norm": 0.00019899260951206088, + "kept_token_count": 572.640625, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894113183021546, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1362199044669976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.4390075683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.66436767578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17480, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30005825459957125, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.12652530893683434, + "grad_norm": 0.00022437056759372354, + "kept_token_count": 572.675, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887923657894134, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.378000085125677e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.55603942871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.43597106933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17490, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.3009186387062073, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.11963315680623055, + "grad_norm": 0.00020348824909888208, + "kept_token_count": 571.9578125, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897692501544952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.996404557256028e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.51817626953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72886962890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17500, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29969067871570587, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.12873073294758797, + "grad_norm": 0.0003056301502510905, + "kept_token_count": 572.7234375, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9899024605751038, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.944368963857414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1056365966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7022705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17510, + "valid_token_count": 817.8109375 + }, + { + "actual_token_mask_ratio": 0.3004670262336731, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.13812071606516838, + "grad_norm": 0.000244409340666607, + "kept_token_count": 572.6671875, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9881301820278168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.636671328626107e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6842803955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1775329589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17520, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.2998403638601303, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.12973841279745102, + "grad_norm": 0.00022148470452520996, + "kept_token_count": 572.8296875, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894493460655213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.121364290767815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.49285583496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0426483154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17530, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2997597008943558, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.13575634509325027, + "grad_norm": 0.00037133877049200237, + "kept_token_count": 572.9828125, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888093292713165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.371377726783976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8023956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0903045654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.30061950981616975, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.12857382819056512, + "grad_norm": 0.00024743328685872257, + "kept_token_count": 571.965625, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989243996143341, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.201584706606809e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4174865722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7525604248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17550, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.29970053434371946, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.12782487720251084, + "grad_norm": 0.00031057081650942564, + "kept_token_count": 572.890625, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887869954109192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3800994171760975e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.04295959472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0692657470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17560, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.30157824158668517, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.12202378660440445, + "grad_norm": 0.0002573076344560832, + "kept_token_count": 571.725, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894443452358246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.123319704376627e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1224029541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3421905517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17570, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.301596000790596, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.13766969591379166, + "grad_norm": 0.0001925878314068541, + "kept_token_count": 571.234375, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894386231899261, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1255544783780354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2246612548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5552001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17580, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.30013218224048616, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.12673384919762612, + "grad_norm": 0.00024547657812945545, + "kept_token_count": 572.6375, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892721354961396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1905927719199096e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.02525024414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.64659729003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17590, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.29941923916339874, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.12597435861825942, + "grad_norm": 0.00027613420388661325, + "kept_token_count": 573.1625, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893114745616913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.175225440121722e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6079864501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8496490478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17600, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.30061450600624084, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.12332646623253822, + "grad_norm": 0.00024226696405094117, + "kept_token_count": 572.3859375, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895819664001465, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.069564274686854e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0541625976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.30077209472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17610, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3010930985212326, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.12269501090049743, + "grad_norm": 0.0002641997707542032, + "kept_token_count": 571.9890625, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889957189559937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2985679829143916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9476776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9586547851562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3004125773906708, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.1315536268055439, + "grad_norm": 0.000270916149020195, + "kept_token_count": 572.6421875, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884139120578765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.525831900537014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.25342712402346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17630, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2998756617307663, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.1310164026916027, + "grad_norm": 0.00024899819982238114, + "kept_token_count": 572.9890625, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989296942949295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.180896175967064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4404235839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.69141845703126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17640, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2990303307771683, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.1269329309463501, + "grad_norm": 0.00020566511375363916, + "kept_token_count": 573.5546875, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893366456031799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.165393947914708e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.95665283203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9584899902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17650, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29910822212696075, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.12338145449757576, + "grad_norm": 0.00027583364862948656, + "kept_token_count": 573.778125, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884065866470337, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.528694262262434e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.79669189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.12061767578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17660, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.2997751712799072, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.1183304361999035, + "grad_norm": 0.0002783545642159879, + "kept_token_count": 572.95625, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0, + "mgd_cos_sim": 0.988858163356781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.352304313215427e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2014221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7064239501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17670, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3013167351484299, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.12630088552832602, + "grad_norm": 0.00018585199723020196, + "kept_token_count": 571.5859375, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889811277389526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.304271533328574e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.19440307617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.98406982421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17680, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.29919158220291137, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.13784391507506372, + "grad_norm": 0.00026051545864902437, + "kept_token_count": 573.6171875, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893039762973785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.178157068963628e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6802947998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4554412841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17690, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.3000940352678299, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.12783995792269706, + "grad_norm": 0.00020768394460901618, + "kept_token_count": 572.6375, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890599250793457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2734794988064095e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9520568847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.764990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17700, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2995464354753494, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.12181814461946487, + "grad_norm": 0.00026298771263100207, + "kept_token_count": 573.0171875, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891241073608399, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2484101868467405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.8483856201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1909973144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17710, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2989963233470917, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.12164857834577561, + "grad_norm": 0.00027114353724755347, + "kept_token_count": 573.784375, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989433640241623, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.127506035729311e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4950836181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.35735778808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17720, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.30110497772693634, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.1368970699608326, + "grad_norm": 0.00018230685964226723, + "kept_token_count": 572.00625, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890665411949158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.270901226846036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.11949462890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5708312988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17730, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2998380124568939, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.11197737157344818, + "grad_norm": 0.000258073239820078, + "kept_token_count": 572.921875, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893028497695923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.17859588196734e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.27903137207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3189239501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17740, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.30064368844032285, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.12260070517659187, + "grad_norm": 0.00024707883130759, + "kept_token_count": 572.45625, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898273527622223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.973709353886079e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1789855957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9722503662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17750, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.3001186728477478, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.11548416689038277, + "grad_norm": 0.0002601323649287224, + "kept_token_count": 572.7671875, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892315030097961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.206463418086059e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8922058105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55784912109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17760, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2991465538740158, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.12374190017580985, + "grad_norm": 0.00025551015278324485, + "kept_token_count": 573.4671875, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893136203289032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1743876863620244e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.92351989746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7825927734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17770, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2996742606163025, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.1256074421107769, + "grad_norm": 0.0003375333617441356, + "kept_token_count": 573.121875, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895026206970214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1005598905030635e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.95916748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1152587890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17780, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.30012622475624084, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.13003353253006936, + "grad_norm": 0.00021995793213136494, + "kept_token_count": 572.6046875, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895049035549164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0996692405315116e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.07057800292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5722259521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17790, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.29986773133277894, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.12355541363358498, + "grad_norm": 0.00025061494670808315, + "kept_token_count": 573.0203125, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890212297439576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.288607888156548e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.17088928222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.19317321777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17800, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.2993392288684845, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.13175645470619202, + "grad_norm": 0.00022797418932896107, + "kept_token_count": 573.309375, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895389974117279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.086347435077187e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6186553955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2173248291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.300965017080307, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.13371991291642188, + "grad_norm": 0.00027908547781407833, + "kept_token_count": 571.796875, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889035880565643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.334553304943256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1042510986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.57737121582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17820, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.2992577314376831, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.12568031772971153, + "grad_norm": 0.00029469767468981445, + "kept_token_count": 573.38125, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893516540527344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.159531890763901e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8289031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0057861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17830, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3003259927034378, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.13351257964968682, + "grad_norm": 0.0002737912873271853, + "kept_token_count": 572.53125, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9876763105392456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.813960949832108e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.13800354003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17840, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.30064484775066375, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.13531856387853622, + "grad_norm": 0.00029272938263602555, + "kept_token_count": 572.0203125, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892486572265625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1997644075308926e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6451477050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.92926635742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17850, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.2999358981847763, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.13074954301118852, + "grad_norm": 0.0001808597007766366, + "kept_token_count": 572.9875, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897147297859192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0176993206841874e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.09498596191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.20350952148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17860, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2995567709207535, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.13416573405265808, + "grad_norm": 0.00019162066746503115, + "kept_token_count": 573.3140625, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895954847335815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0642829117132354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3957489013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.84385070800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17870, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29953572154045105, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.11683535873889923, + "grad_norm": 0.00032905713305808604, + "kept_token_count": 573.0796875, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890589058399201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.273883023415692e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2199188232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3404541015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17880, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2991532951593399, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.12752960473299027, + "grad_norm": 0.00021003420988563448, + "kept_token_count": 573.3984375, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891947150230408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2208359445794486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8530639648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6182342529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17890, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.2996596425771713, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.12155893296003342, + "grad_norm": 0.00019674532813951373, + "kept_token_count": 573.2234375, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989945775270462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9274516529985705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2053985595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0822235107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17900, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.3008476674556732, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.12526329681277276, + "grad_norm": 0.0003134076250717044, + "kept_token_count": 572.1984375, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895283341407776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0905111382016913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.28778076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5136444091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17910, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.301009464263916, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.12844648733735084, + "grad_norm": 0.00023296191648114473, + "kept_token_count": 571.6296875, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891051411628723, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.255820822436362e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.62841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8054138183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17920, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.30055729746818544, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.11889837831258773, + "grad_norm": 0.0002538431726861745, + "kept_token_count": 572.259375, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895937860012054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0649429138284177e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.36037292480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4448181152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17930, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.29987800419330596, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.12767234593629836, + "grad_norm": 0.000328498164890334, + "kept_token_count": 572.809375, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887936890125275, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.377483564894646e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.13794860839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3365509033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17940, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2999291032552719, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.13105615824460984, + "grad_norm": 0.00026802567299455404, + "kept_token_count": 572.76875, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892742276191712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1897724076989104e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.84239807128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6584869384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17950, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29932514727115633, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.1266784854233265, + "grad_norm": 0.0002775931789074093, + "kept_token_count": 573.390625, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884781002998352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.5007602602709083e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9144989013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.97833557128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17960, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.30006741881370547, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.12748862728476523, + "grad_norm": 0.0002909134782385081, + "kept_token_count": 572.6640625, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889645636081695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.310733493184671e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.49317932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34911193847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.299441933631897, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.12180742621421814, + "grad_norm": 0.00024673555162735283, + "kept_token_count": 573.296875, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891377627849579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2430827306816354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.96213989257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.14293518066404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17980, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.30052571892738345, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.12413030564785003, + "grad_norm": 0.0004648107278626412, + "kept_token_count": 572.4765625, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894918441772461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.104771360289305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.63780822753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8354187011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17990, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2995469719171524, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.12423244118690491, + "grad_norm": 0.00023716066789347678, + "kept_token_count": 573.4515625, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895197808742523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.093855131941382e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.21226806640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8019989013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18000, + "valid_token_count": 818.6828125 + }, + { + "actual_token_mask_ratio": 0.2997529894113541, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.12013414055109024, + "grad_norm": 0.00022060004994273186, + "kept_token_count": 572.9390625, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894909977912902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.10509979701601e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.64051513671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9173889160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2997299790382385, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.12950782626867294, + "grad_norm": 0.00023984044673852623, + "kept_token_count": 573.053125, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893435359001159, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.162700206506997e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4358673095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.18287658691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18020, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2997657239437103, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.12546920999884606, + "grad_norm": 0.00027730356669053435, + "kept_token_count": 573.0546875, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890962243080139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2593117905198595e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.55975646972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2877258300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18030, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30083417892456055, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.13541694432497026, + "grad_norm": 0.0002010515599977225, + "kept_token_count": 571.9875, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989777535200119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.993172686023172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7255035400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.00299682617185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18040, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.30050545930862427, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.13649413660168647, + "grad_norm": 0.0002079314144793898, + "kept_token_count": 572.375, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9900496900081635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8868615229148416e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.51732482910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.33656005859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18050, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3001114547252655, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.12834324762225152, + "grad_norm": 0.0003642937808763236, + "kept_token_count": 572.9, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888744711875915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3459311200422235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1577575683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.27121887207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18060, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.29992454349994657, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.13495358377695083, + "grad_norm": 0.00018346219439990819, + "kept_token_count": 572.9, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897123634815216, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.018632826046087e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5607604980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77711791992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18070, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3004183202981949, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.1284647010266781, + "grad_norm": 0.00022822328901384026, + "kept_token_count": 572.4328125, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889510691165924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.316001432016492e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3491638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0292602539063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18080, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.29955274164676665, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.1270662374794483, + "grad_norm": 0.00026783772045746446, + "kept_token_count": 573.2109375, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893609046936035, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1559121018508446e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4619537353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7156982421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18090, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2995517522096634, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.1181240014731884, + "grad_norm": 0.00022530692513100803, + "kept_token_count": 573.09375, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889328956604004, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.323106186348014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.93165283203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1891723632813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18100, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29972605109214784, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.1274666056036949, + "grad_norm": 0.00023568376491311938, + "kept_token_count": 572.93125, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890460550785065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.278903761587571e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8973724365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8344329833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18110, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.29966401755809785, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.1240703858435154, + "grad_norm": 0.00026873088791035116, + "kept_token_count": 573.1234375, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890980303287507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2586006384226496e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.33197631835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3772033691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3007445722818375, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.13067716285586356, + "grad_norm": 0.0002914285287261009, + "kept_token_count": 572.303125, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9884965360164643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.493555243243463e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.52874145507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3650207519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18130, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.30003869235515596, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.13241473734378814, + "grad_norm": 0.00021314249897841364, + "kept_token_count": 572.7359375, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898939728736877, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.947690856875852e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4558990478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.34694519042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18140, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.30079585313796997, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.12406142428517342, + "grad_norm": 0.0002649984380695969, + "kept_token_count": 572.0296875, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9903155863285065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7829937355127186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2838195800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.69833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18150, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3007539719343185, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.13005366921424866, + "grad_norm": 0.0002474893117323518, + "kept_token_count": 571.99375, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887820899486541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.382016049930826e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.75078735351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8917510986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18160, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2994346559047699, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.12586845457553864, + "grad_norm": 0.00020164847956039011, + "kept_token_count": 573.2359375, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893260836601258, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.169513558736071e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0591705322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.76813049316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18170, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29839090406894686, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.1258068025112152, + "grad_norm": 0.0001956033956957981, + "kept_token_count": 574.0375, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989122885465622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.248896439094096e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.12181091308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5384216308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18180, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.3007416844367981, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.13324921131134032, + "grad_norm": 0.00027829327154904604, + "kept_token_count": 571.959375, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894093096256256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.137012583669275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0726715087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.38798828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18190, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.3004562586545944, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.12213154137134552, + "grad_norm": 0.00025045647635124624, + "kept_token_count": 572.6703125, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896352708339691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.04874994273996e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.67711486816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7277770996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18200, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.30023660957813264, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.12267563045024872, + "grad_norm": 0.0002899549435824156, + "kept_token_count": 572.5875, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891085922718048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2544737880234607e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.76930236816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1709411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18210, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.30096185207366943, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.13558573722839357, + "grad_norm": 0.00023973021598067135, + "kept_token_count": 572.1203125, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891130447387695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.252734943293035e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8587677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8794006347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18220, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.300874137878418, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.12373347580432892, + "grad_norm": 0.0002495921798981726, + "kept_token_count": 571.840625, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891137719154358, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.252455946698319e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0716857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31610107421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18230, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.30136300921440123, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.13260112032294274, + "grad_norm": 0.00024160715111065656, + "kept_token_count": 571.4578125, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893281400203705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1687166230985895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.86549682617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2833770751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18240, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.3000102907419205, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.12650241404771806, + "grad_norm": 0.00020309169485699385, + "kept_token_count": 572.5875, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898223519325257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.975662402808666e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7722137451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9490997314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29983898997306824, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.12446819916367531, + "grad_norm": 0.00020252837566658854, + "kept_token_count": 572.8765625, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893106102943421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.175563299213536e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5239624023437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.95636596679685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18260, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2988165646791458, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.1372004672884941, + "grad_norm": 0.00021747020946349949, + "kept_token_count": 573.7515625, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888485848903656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.35604033555137e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.87843017578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.23194274902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18270, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2998599112033844, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.12001513317227364, + "grad_norm": 0.00020074793428648263, + "kept_token_count": 572.9359375, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9899776458740235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.914995431841817e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9513244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6257019042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18280, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30004697740077974, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.12386018112301826, + "grad_norm": 0.0002553212398197502, + "kept_token_count": 572.5375, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9892039358615875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.217230562062468e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.09851989746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5293365478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18290, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.30096069276332854, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.12717448845505713, + "grad_norm": 0.00029590423218905926, + "kept_token_count": 571.928125, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895637452602386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0766829988569954e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5867065429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4551025390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18300, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.29993765950202944, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.12652973905205728, + "grad_norm": 0.00021869296324439347, + "kept_token_count": 572.8515625, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893946468830108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.142732832406182e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7465454101563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30042973458766936, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.12116918116807937, + "grad_norm": 0.0002110506611643359, + "kept_token_count": 572.4515625, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888673067092896, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.348726761236321e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.39959411621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7438629150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18320, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.300104358792305, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.13574930429458618, + "grad_norm": 0.00023559197143185884, + "kept_token_count": 572.5921875, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887339174747467, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.400832622195594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.90928955078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5480194091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18330, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.3004365384578705, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.1331088773906231, + "grad_norm": 0.00028042730991728604, + "kept_token_count": 572.5765625, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882602870464325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.58584057923872e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.78119506835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2191589355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18340, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.2998038411140442, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.12732605263590813, + "grad_norm": 0.00020949187455698848, + "kept_token_count": 572.93125, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898627161979675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.959892128477804e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2391021728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8630889892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18350, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3002352356910706, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.12477239370346069, + "grad_norm": 0.00035484295221976936, + "kept_token_count": 572.659375, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894010365009308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1402371061849406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5489837646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4589508056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18360, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29957696199417116, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.13579974621534346, + "grad_norm": 0.00023523486743215472, + "kept_token_count": 573.0765625, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896322369575501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.049923954880796e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7092041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.49256591796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18370, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2994198679924011, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.13422973230481147, + "grad_norm": 0.00023683778999838978, + "kept_token_count": 573.346875, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895141899585724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0960367914522064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6949951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7752197265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18380, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2989701211452484, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.1285938046872616, + "grad_norm": 0.00023595569655299187, + "kept_token_count": 573.8421875, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897698938846589, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.996150044258684e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.44158325195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5604583740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18390, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.3002521723508835, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.1360912598669529, + "grad_norm": 0.00026798894396051764, + "kept_token_count": 572.646875, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989478713274002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.109900837647729e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.03026123046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.09156494140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18400, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.29952155351638793, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.12938174828886986, + "grad_norm": 0.0002504322328604758, + "kept_token_count": 573.5546875, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894589960575104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.117600728932302e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3339050292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0942413330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18410, + "valid_token_count": 818.8 + }, + { + "actual_token_mask_ratio": 0.3005290120840073, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.1229749120771885, + "grad_norm": 0.00022618984803557396, + "kept_token_count": 572.3765625, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9901725232601166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.838882112177089e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.309912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.77911987304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18420, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.30068560540676115, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.13118760511279107, + "grad_norm": 0.00020279861928429455, + "kept_token_count": 572.3, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895487487316131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0825377800501884e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8132659912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5165130615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18430, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.3009671688079834, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.121256622672081, + "grad_norm": 0.00028369182837195694, + "kept_token_count": 572.1265625, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888112843036652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.370609676698223e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.30860290527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18440, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.29931117296218873, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.1280316412448883, + "grad_norm": 0.0002669255482032895, + "kept_token_count": 573.415625, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891865491867066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2240245602442886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2890228271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.400341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18450, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2993540793657303, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.13012754321098327, + "grad_norm": 0.0002935949887614697, + "kept_token_count": 573.296875, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989734822511673, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.009854601463303e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5276397705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.78277282714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18460, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.300593426823616, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.12014327272772789, + "grad_norm": 0.00020588957704603672, + "kept_token_count": 572.3078125, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893508911132812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.159823292866349e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5360412597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.48801574707034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18470, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3001758217811584, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.1248105637729168, + "grad_norm": 0.00021560447930824012, + "kept_token_count": 572.840625, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895139396190643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.096136399311945e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.15159606933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6965637207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18480, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.30038235187530515, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.1305658385157585, + "grad_norm": 0.00019197689834982157, + "kept_token_count": 572.421875, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9889193117618561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.328410868765786e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.42812194824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9960968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.298717799782753, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.11808786541223526, + "grad_norm": 0.0001849538239184767, + "kept_token_count": 573.903125, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9899387240409852, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9302085133385846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7432434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.74302673339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.2990844249725342, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.13205335065722465, + "grad_norm": 0.00022938093752600253, + "kept_token_count": 573.6921875, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895946145057678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.064627501065843e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3548889160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.231298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18510, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.29974194765090945, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.12243112474679947, + "grad_norm": 0.0002514219668228179, + "kept_token_count": 573.25625, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895106196403504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.097433884453494e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.79148254394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.70811462402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18520, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.30054864287376404, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.1249624215066433, + "grad_norm": 0.00024137605214491487, + "kept_token_count": 572.3890625, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9899686992168426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.918492147931829e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.34153137207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2040771484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18530, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.300031316280365, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.12936320379376412, + "grad_norm": 0.00019767369667533785, + "kept_token_count": 572.9109375, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894452452659607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.122968784940895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.6101593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.91809997558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18540, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.30136236548423767, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.134368996322155, + "grad_norm": 0.00025679171085357666, + "kept_token_count": 571.684375, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896614849567413, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.038499573653098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4669586181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44829711914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18550, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3006177097558975, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.1274067647755146, + "grad_norm": 0.0003315032518003136, + "kept_token_count": 572.1, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894390642642975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1253811650676654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2936309814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.143115234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18560, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.29930253624916076, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.1257045641541481, + "grad_norm": 0.0002813475439324975, + "kept_token_count": 573.046875, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9890365958213806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2826024946407414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2837860107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9946594238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18570, + "valid_token_count": 817.8171875 + }, + { + "actual_token_mask_ratio": 0.3002958208322525, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.12143905758857727, + "grad_norm": 0.0002251038240501657, + "kept_token_count": 572.73125, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9888090133666992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3715012725442645e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.47158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.97105102539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18580, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.3012020856142044, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.12405378594994546, + "grad_norm": 0.0003104671777691692, + "kept_token_count": 571.8984375, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893288493156434, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1684328243718484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.16224670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.15855102539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18590, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.2998279303312302, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.12588493824005126, + "grad_norm": 0.000241429777815938, + "kept_token_count": 572.7578125, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9898216128349304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.975955187343061e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.62804260253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.15468139648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18600, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.30042134821414945, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.12440471276640892, + "grad_norm": 0.0002288338728249073, + "kept_token_count": 572.6296875, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897654592990875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.997883177362382e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3501831054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4993103027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18610, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.30143981575965884, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.12241309434175492, + "grad_norm": 0.0002614237309899181, + "kept_token_count": 571.440625, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989327210187912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.169076564721763e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.10040893554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5221008300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18620, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.29831532537937167, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.12608083710074425, + "grad_norm": 0.00019106947002001107, + "kept_token_count": 574.3265625, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894508957862854, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.120769735891372e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.60301208496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.20379333496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18630, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2994930028915405, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.12048278599977494, + "grad_norm": 0.0002518284891266376, + "kept_token_count": 573.428125, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891523241996765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.237394059600774e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4239562988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.5108306884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18640, + "valid_token_count": 818.5875 + }, + { + "actual_token_mask_ratio": 0.29949679374694826, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.1244027353823185, + "grad_norm": 0.00033738091588020325, + "kept_token_count": 573.2078125, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894482016563415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.121812999073882e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7723876953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9614593505859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2996097683906555, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.12581088095903398, + "grad_norm": 0.00021098266006447375, + "kept_token_count": 573.2, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897624433040619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9990574441617356e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.72449645996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.43532104492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18660, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29958856403827666, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.11989066749811172, + "grad_norm": 0.00028311871574260294, + "kept_token_count": 573.090625, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891097068786621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2540438153082504e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.43792419433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.90321044921876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18670, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3000821262598038, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.12556612864136696, + "grad_norm": 0.0002547292097005993, + "kept_token_count": 573.0140625, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9897370338439941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.008994728792459e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4686737060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.08704833984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18680, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.29970458447933196, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.12612186148762702, + "grad_norm": 0.0002173113462049514, + "kept_token_count": 572.80625, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896071016788482, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.059738530486357e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5466003417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.45509948730466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18690, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.3002864420413971, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.12457854300737381, + "grad_norm": 0.00022939797781873494, + "kept_token_count": 572.4890625, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893144786357879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1740512097021563e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2321319580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4381042480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18700, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.30012045204639437, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.11874056756496429, + "grad_norm": 0.0005513724172487855, + "kept_token_count": 572.7765625, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0, + "mgd_cos_sim": 0.989151531457901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2377039790153503e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.69556579589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.96845092773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18710, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2989745318889618, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.1312235876917839, + "grad_norm": 0.0003048714133910835, + "kept_token_count": 573.440625, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9891434609889984, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.240856032993179e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.42954711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4328948974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18720, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.3003129482269287, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.12401180118322372, + "grad_norm": 0.00024160374596249312, + "kept_token_count": 572.6765625, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9882944107055665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.57251924672164e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9228790283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.53993835449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18730, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.29997296035289767, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.13629538118839263, + "grad_norm": 0.0002576670376583934, + "kept_token_count": 573.18125, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9893825352191925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1474694808130155e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4828369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.29278259277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18740, + "valid_token_count": 818.8015625 + }, + { + "actual_token_mask_ratio": 0.30041739344596863, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.13269643113017082, + "grad_norm": 0.00017603639571461827, + "kept_token_count": 572.475, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9900821685791016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.87417330784956e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.49608459472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79918212890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18750, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3001295953989029, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.1254948101937771, + "grad_norm": 0.0002698241441976279, + "kept_token_count": 572.725, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9895892202854156, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.066728834004607e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7532928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5586853027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18760, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29964591562747955, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.12490683197975158, + "grad_norm": 0.00029540309333242476, + "kept_token_count": 573.15, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9902455031871795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8103670522104946e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7814971923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.05360107421876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18770, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2998201549053192, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.12306661829352379, + "grad_norm": 0.0001888579863589257, + "kept_token_count": 573.0140625, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9896277904510498, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0516682020097505e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.46312255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4853759765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18780, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.29952298998832705, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.1315868690609932, + "grad_norm": 0.00022289849584922194, + "kept_token_count": 573.0796875, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887428104877471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.397360135044437e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.66468505859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9913848876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.3004795163869858, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.1399304524064064, + "grad_norm": 0.00020029717416036874, + "kept_token_count": 572.715625, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9887898206710816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.378998237370979e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.88872985839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.312109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18800, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.3002462834119797, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.1244466908276081, + "grad_norm": 0.0003080389287788421, + "kept_token_count": 572.6484375, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9894489467144012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.121528472751379e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.81463928222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.26911010742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18810, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.299655357003212, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.12769311368465425, + "grad_norm": 0.00020209424837958068, + "kept_token_count": 573.2625, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9899464428424836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.927191319235135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0881011962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2647308349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18820, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.3005582958459854, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.13199393078684807, + "grad_norm": 0.00023626875190529972, + "kept_token_count": 572.209375, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0, + "mgd_cos_sim": 0.9900532603263855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8854594822623764e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.53387756347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.98109436035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18830, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30013103485107423, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.1308520630002022, + "grad_norm": 0.0002504016156308353, + "kept_token_count": 572.6609375, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894231796264649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.131589739699848e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5993591308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5033233642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18840, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29805195033550264, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.12900158315896987, + "grad_norm": 0.0002775966131594032, + "kept_token_count": 574.4984375, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894713997840882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.11274959333241e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5939056396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.94660034179685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18850, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.3001684874296188, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.12959966138005258, + "grad_norm": 0.00020640605362132192, + "kept_token_count": 572.478125, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896105349063873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0584044472780076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.84974670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0074432373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18860, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.3004878520965576, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.1317007400095463, + "grad_norm": 0.0002638926962390542, + "kept_token_count": 572.465625, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899955630302429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9080029455362816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7727386474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.95521240234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18870, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.29922651648521426, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.12438989877700805, + "grad_norm": 0.00026507111033424735, + "kept_token_count": 573.515625, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895631432533264, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.076912737218663e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.48460083007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9985046386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18880, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3007744073867798, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.1289919525384903, + "grad_norm": 0.00021865488088224083, + "kept_token_count": 571.9609375, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902280867099762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8171676715137436e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1538482666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.09266662597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18890, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.3002562642097473, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.12858810499310494, + "grad_norm": 0.0001943587267305702, + "kept_token_count": 572.6171875, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891827285289765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.225516677252017e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4300231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4990264892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18900, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3008156269788742, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.12168717384338379, + "grad_norm": 0.00022065237862989306, + "kept_token_count": 572.103125, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897967755794526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.985655366705032e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0697082519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7696472167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18910, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29994118213653564, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.12557066455483437, + "grad_norm": 0.00020503223640844226, + "kept_token_count": 572.8953125, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894846677780151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.107570384803694e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0318572998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0403686523438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18920, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30093042850494384, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.12387162670493126, + "grad_norm": 0.0003041262098122388, + "kept_token_count": 572.0046875, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889849066734314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.302791894588154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2016326904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.10202941894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18930, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.30066738128662107, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.12050574570894242, + "grad_norm": 0.0003379214322194457, + "kept_token_count": 572.296875, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892407894134522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.202844575047493e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.98780517578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42291259765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18940, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.3002647966146469, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.12021776810288429, + "grad_norm": 0.00026233421522192657, + "kept_token_count": 572.5953125, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9886830449104309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.420698351168539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.52125549316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79273986816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18950, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.3010231852531433, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.1269820399582386, + "grad_norm": 0.00026830376009456813, + "kept_token_count": 571.921875, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9890992045402527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2581379966577514e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9635803222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.50928955078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18960, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29975603222846986, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.12993597313761712, + "grad_norm": 0.0002080648555420339, + "kept_token_count": 572.7765625, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989482843875885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.108287230337737e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.457763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9594757080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18970, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.3008880317211151, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.12094787582755089, + "grad_norm": 0.0002706947852857411, + "kept_token_count": 572.05625, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897941410541534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.986682568211108e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.04407653808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.84316101074216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18980, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.29992925226688383, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.1230042226612568, + "grad_norm": 0.0003661655937321484, + "kept_token_count": 572.9796875, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892671406269073, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.19254633015953e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8356567382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9781951904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18990, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2994818210601807, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.12986075058579444, + "grad_norm": 0.000473325839266181, + "kept_token_count": 573.246875, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889346241950989, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.322431614127709e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.39813842773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.828564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19000, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2998057812452316, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.12743528187274933, + "grad_norm": 0.0002632702235132456, + "kept_token_count": 572.9421875, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897898614406586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9883539102447685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8336242675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9700866699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19010, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3000195652246475, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.1211695559322834, + "grad_norm": 0.00029931662720628083, + "kept_token_count": 572.640625, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894912242889404, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1050108848139644e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.32325439453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.66648864746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19020, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30055617094039916, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.12785218954086303, + "grad_norm": 0.0002546764735598117, + "kept_token_count": 572.3984375, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893777668476105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1493259777780624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6259033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8257171630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19030, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2991628676652908, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.14014039039611817, + "grad_norm": 0.00024900652351789176, + "kept_token_count": 573.71875, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898221850395202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.975736144639086e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.84502868652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.86888732910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19040, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.30065037608146666, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.12754147946834565, + "grad_norm": 0.00024089829821605235, + "kept_token_count": 572.109375, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889470100402832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3175957034691234e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.50827941894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7802337646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19050, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3006783753633499, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.1312842383980751, + "grad_norm": 0.0002468242309987545, + "kept_token_count": 572.1765625, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893972396850585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1417164902668445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3054931640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8333374023438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19060, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.300822451710701, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.13069090023636817, + "grad_norm": 0.00019847328076139092, + "kept_token_count": 572.2328125, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891290307044983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.246492353559006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.21295776367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.82077941894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19070, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.30089311599731444, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.12604573592543603, + "grad_norm": 0.00020941910042893142, + "kept_token_count": 572.05625, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889989495277405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.297306622902397e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.65208435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8509094238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19080, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.29988413155078886, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.1304058887064457, + "grad_norm": 0.0002317040489288047, + "kept_token_count": 572.7625, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989254105091095, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.197638991172425e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6411651611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82305297851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19090, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.29943621456623076, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.1257575884461403, + "grad_norm": 0.00037976622115820646, + "kept_token_count": 573.2609375, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896165192127228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.056071556988172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3106658935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.84601135253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19100, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.3002964615821838, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.14436333104968072, + "grad_norm": 0.0001696277759037912, + "kept_token_count": 572.7484375, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903931319713593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7527031054196414e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4724487304687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7051727294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19110, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.3007372498512268, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.125634153932333, + "grad_norm": 0.00026440693181939423, + "kept_token_count": 571.925, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893303513526917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1678502748254685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8074676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.27900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19120, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.2989876925945282, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.13355481252074242, + "grad_norm": 0.0002506504242774099, + "kept_token_count": 573.89375, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892901599407196, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1835531374090354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6677276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.06043395996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19130, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.29913963079452516, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.13175787031650543, + "grad_norm": 0.00017117836978286505, + "kept_token_count": 573.3109375, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9890668630599976, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.270772842573933e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.56528625488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.550146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19140, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.3005656063556671, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.12858858630061148, + "grad_norm": 0.00025691691553220153, + "kept_token_count": 572.2921875, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893896400928497, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1446895193075764e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.31337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3501892089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19150, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2997945427894592, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.1318069763481617, + "grad_norm": 0.00020552212663460523, + "kept_token_count": 572.9109375, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9887121975421905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.409314424265176e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0683898925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.95953979492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19160, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2998198658227921, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.12107041776180268, + "grad_norm": 0.00026686853379942477, + "kept_token_count": 573.2359375, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898391366004944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.969099452660885e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7422271728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5076080322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19170, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.29981037974357605, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.12731628194451333, + "grad_norm": 0.00020025281992275268, + "kept_token_count": 573.021875, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902640044689178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.803148356382735e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7010925292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.70908203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19180, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2995106607675552, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.12775445356965065, + "grad_norm": 0.00023353530559688807, + "kept_token_count": 573.33125, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893399536609649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.164093552390113e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.49437255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8899444580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19190, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.3008293181657791, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.13811373412609101, + "grad_norm": 0.00026829077978618443, + "kept_token_count": 572.0125, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899352669715882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9315560206887314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.7405029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3332000732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19200, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30149235129356383, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.12954852655529975, + "grad_norm": 0.00021377374650910497, + "kept_token_count": 571.3484375, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897490859031677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.004282236564904e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0454345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6040832519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19210, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.29944791495800016, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.12675287574529648, + "grad_norm": 0.00018487167835701257, + "kept_token_count": 573.3328125, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896282076835632, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.051506221003365e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2710418701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7170013427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19220, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.2990662306547165, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.13417647629976273, + "grad_norm": 0.0002811025769915432, + "kept_token_count": 573.25, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892743468284607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.189730134385172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1216674804688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4491882324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19230, + "valid_token_count": 817.8375 + }, + { + "actual_token_mask_ratio": 0.29941032230854037, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.13098100051283837, + "grad_norm": 0.000259383930824697, + "kept_token_count": 573.08125, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900608956813812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8824809598736466e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7366668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.37803039550784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19240, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.3000344842672348, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.1413610264658928, + "grad_norm": 0.00019837924628518522, + "kept_token_count": 572.8421875, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901263773441314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8568982563447205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.89430541992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1715148925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19250, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.300236240029335, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.1355721488595009, + "grad_norm": 0.0002170677180401981, + "kept_token_count": 572.6375, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899341404438019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.93199639802333e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.17942810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7833221435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19260, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.30106728374958036, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.11390398070216179, + "grad_norm": 0.00020515648066066206, + "kept_token_count": 572.1328125, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899171531200409, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.938632180506829e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.556298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01157531738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19270, + "valid_token_count": 818.584375 + }, + { + "actual_token_mask_ratio": 0.2999540537595749, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.14127939492464064, + "grad_norm": 0.00028450123500078917, + "kept_token_count": 572.5109375, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894365310668946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.126368621655274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3248229980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.692333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19280, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.29972548484802247, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.1308731257915497, + "grad_norm": 0.00022037921007722616, + "kept_token_count": 572.925, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892832338809967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.186254627711605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.85150451660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.65821533203126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19290, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.3000645786523819, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.13337289467453955, + "grad_norm": 0.00022067883401177824, + "kept_token_count": 572.6796875, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989084005355835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2640797983040105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6813629150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4335479736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19300, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2990126758813858, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.13189219012856485, + "grad_norm": 0.00022016101866029203, + "kept_token_count": 573.7390625, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894229292869567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.131682544539217e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.31078186035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8281524658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30002239942550657, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.13067632019519806, + "grad_norm": 0.00022410735255107284, + "kept_token_count": 572.696875, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895819842815399, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0695632560527886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5086181640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3957550048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19320, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.29957703649997713, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.12429117858409881, + "grad_norm": 0.00020714630954898894, + "kept_token_count": 573.2015625, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908602893352508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5702201421372594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0416320800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44097900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19330, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3004716277122498, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.12951211258769035, + "grad_norm": 0.0003336168883834034, + "kept_token_count": 572.4328125, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901902377605438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.831963695120066e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.79523010253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0318634033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19340, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3004477500915527, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.11907447502017021, + "grad_norm": 0.0001763209729688242, + "kept_token_count": 572.403125, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895697593688965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0743398858467114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.23609619140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.95238037109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.3002635180950165, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.12493918016552925, + "grad_norm": 0.0002494528889656067, + "kept_token_count": 572.5, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895643055438995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0764690857031384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3017608642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.93846435546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19360, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2990808427333832, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.1303764335811138, + "grad_norm": 0.0002093879011226818, + "kept_token_count": 573.5015625, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896698296070099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.035242636746261e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3087982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.42828674316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19370, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.3009848177433014, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.13069468811154367, + "grad_norm": 0.00016999115177895874, + "kept_token_count": 572.05, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895287334918976, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0903575063566676e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0544708251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6670379638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19380, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.300506255030632, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.12831371724605561, + "grad_norm": 0.00024337701324839145, + "kept_token_count": 572.5265625, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898367106914521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.970050638599787e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8230743408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.640478515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.3005289971828461, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.13466460704803468, + "grad_norm": 0.00020081082766409963, + "kept_token_count": 572.4796875, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897778391838074, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.99304619350005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41843872070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0489776611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19400, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30025698244571686, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.12014774084091187, + "grad_norm": 0.00020536470401566476, + "kept_token_count": 572.2484375, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907346308231354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.619305134634487e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3249450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9364776611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19410, + "valid_token_count": 817.803125 + }, + { + "actual_token_mask_ratio": 0.3001830756664276, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.1265472263097763, + "grad_norm": 0.00023307646915782243, + "kept_token_count": 572.396875, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893624544143677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.155310525675304e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9239898681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7864990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19420, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.3007208168506622, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.12021407708525658, + "grad_norm": 0.00032848480623215437, + "kept_token_count": 571.9921875, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902281939983368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.817133620032109e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.41286315917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.70404663085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19430, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.30011328756809236, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.13764014169573785, + "grad_norm": 0.0001845070073613897, + "kept_token_count": 572.6265625, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0, + "mgd_cos_sim": 0.988954770565033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3145580275449905e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.625244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.76817016601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19440, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2991507798433304, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.11714223101735115, + "grad_norm": 0.0004464926605578512, + "kept_token_count": 573.5234375, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899153292179108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.93933860323159e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7615600585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19450, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.300417360663414, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.13364391922950744, + "grad_norm": 0.00021602083870675415, + "kept_token_count": 572.1984375, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989909154176712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.941760514862836e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7273315429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5114440917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19460, + "valid_token_count": 817.90625 + }, + { + "actual_token_mask_ratio": 0.2995433956384659, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.1313044473528862, + "grad_norm": 0.0003108027158305049, + "kept_token_count": 573.09375, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896572887897491, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0401377555099315e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.79457702636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4768310546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19470, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30042449831962587, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.1319177582859993, + "grad_norm": 0.00026876945048570633, + "kept_token_count": 572.3171875, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897234559059143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.014296391687822e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.66636657714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.18096923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3005336344242096, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.12991748079657556, + "grad_norm": 0.00021289206051733345, + "kept_token_count": 572.6296875, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896362960338593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0483387783751824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1495635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0682647705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19490, + "valid_token_count": 818.6609375 + }, + { + "actual_token_mask_ratio": 0.29944336116313935, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.1319733515381813, + "grad_norm": 0.0001820102334022522, + "kept_token_count": 573.2421875, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893912851810456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.144047998124734e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4085174560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2685577392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19500, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3011795312166214, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.13426322862505913, + "grad_norm": 0.00018687709234654903, + "kept_token_count": 571.753125, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900475323200226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.887703896907624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4884063720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6522552490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19510, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.30027438402175904, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.13862633779644967, + "grad_norm": 0.00023720170429442078, + "kept_token_count": 572.3765625, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895031034946442, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.10037420806475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3234619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6762268066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19520, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.3000355243682861, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.13636651560664176, + "grad_norm": 0.0003177207545377314, + "kept_token_count": 572.75625, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9887273132801055, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.4034069287590685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7476104736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9001037597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19530, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30023232400417327, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.1337045207619667, + "grad_norm": 0.00030839117243885994, + "kept_token_count": 572.715625, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9890060305595398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2945385939674455e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.30861511230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.02763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19540, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.3008199393749237, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.12773474901914597, + "grad_norm": 0.00023101719853002578, + "kept_token_count": 572.0640625, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897697150707245, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9962268419913014e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5681182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7952911376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19550, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2996847242116928, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.12505074888467788, + "grad_norm": 0.00021505630866158754, + "kept_token_count": 573.1828125, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893505334854126, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1599653923185545e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6572784423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2923553466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19560, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.2995530992746353, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.12912260219454766, + "grad_norm": 0.0002304490772075951, + "kept_token_count": 573.340625, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896793007850647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.031540229334496e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.35231323242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3441619873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19570, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30041300058364867, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.13266648277640342, + "grad_norm": 0.0002428905136184767, + "kept_token_count": 572.39375, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892795801162719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.187681806797627e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.42412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5460540771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19580, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29981496930122375, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.12185680195689201, + "grad_norm": 0.00021085559274069965, + "kept_token_count": 573.1421875, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899697780609131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9180809835670516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7989105224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1285858154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19590, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.3003352522850037, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.12545055896043777, + "grad_norm": 0.0002051269548246637, + "kept_token_count": 572.6421875, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900934100151062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8697788113495335e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.482080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0163940429687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19600, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.29875686168670657, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.13082300275564193, + "grad_norm": 0.0002562966838013381, + "kept_token_count": 573.8890625, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899465799331665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9271376954275185e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.42138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9290802001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19610, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3001988887786865, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.1259879246354103, + "grad_norm": 0.00025525264209136367, + "kept_token_count": 572.540625, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903174698352813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7822631384187844e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.90984802246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4406677246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19620, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2994386851787567, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.13706770837306975, + "grad_norm": 0.00024296905030496418, + "kept_token_count": 573.3140625, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894752323627471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.111254929739516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3107543945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28891906738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19630, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.3005881756544113, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.13613251298666001, + "grad_norm": 0.00025884094065986574, + "kept_token_count": 572.3, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894952416419983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1034432433662006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.59076538085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.54677124023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19640, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2997665047645569, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.13088670521974563, + "grad_norm": 0.00022242330305743963, + "kept_token_count": 572.9171875, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895094096660614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0979067125590515e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1955902099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.00160827636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19650, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2992303937673569, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.13824633583426477, + "grad_norm": 0.0001678478583926335, + "kept_token_count": 573.3625, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892995536327363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1798827805905606e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.61167297363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.60263977050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19660, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.30027752816677095, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.12762898281216623, + "grad_norm": 0.0002507820026949048, + "kept_token_count": 572.6734375, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892578363418579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.196175541437697e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.61749267578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3364593505859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19670, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30104851722717285, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.13299267888069152, + "grad_norm": 0.00022620445815846324, + "kept_token_count": 571.7546875, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895910918712616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0659992737346326e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5767547607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8942413330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19680, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2999074339866638, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.13774716705083848, + "grad_norm": 0.00028481543995440006, + "kept_token_count": 572.996875, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893236041069031, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1704842442413795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.390869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.77086181640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19690, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.29964938163757326, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.1266497127711773, + "grad_norm": 0.00021088949870318174, + "kept_token_count": 573.0328125, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899218857288361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.93678252294194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.00517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.87660827636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19700, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3012499392032623, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.12573178187012674, + "grad_norm": 0.00020206654153298587, + "kept_token_count": 571.684375, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897597670555115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.000107255706098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.15602722167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.41712646484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19710, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.30058465003967283, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.13047000542283058, + "grad_norm": 0.00025482915225438774, + "kept_token_count": 572.5015625, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907189667224884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.625421613833169e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8631164550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.19871215820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19720, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2996441930532455, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.13613269552588464, + "grad_norm": 0.0002213095867773518, + "kept_token_count": 572.865625, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901285290718078, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.856062230624957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.79535522460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4915740966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19730, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.29995899796485903, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.12931793928146362, + "grad_norm": 0.0002834665065165609, + "kept_token_count": 572.584375, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902043819427491, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8264331305981616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5940399169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.310888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19740, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.29990485608577727, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.11807401999831199, + "grad_norm": 0.00019141464144922793, + "kept_token_count": 572.8625, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891223430633544, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.2491058411542324e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.30601196289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.10700073242185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19750, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.3013525426387787, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.12211339995265007, + "grad_norm": 0.0002937029639724642, + "kept_token_count": 571.5984375, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892447650432586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.201285373710562e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0137237548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.57208862304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19760, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29930721819400785, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.1229502409696579, + "grad_norm": 0.00019907354726456106, + "kept_token_count": 573.30625, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900938510894776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.869612846756354e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6733123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.08274536132814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19770, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2996721386909485, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.13487670123577117, + "grad_norm": 0.00021540444868151098, + "kept_token_count": 572.9828125, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901719689369202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.839091368718073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.85682678222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.34722900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19780, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.300733807682991, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.13314411118626596, + "grad_norm": 0.00026594792143441737, + "kept_token_count": 572.3078125, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898907780647278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9489338450948705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1379150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6857025146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19790, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.29992568790912627, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.13280325382947922, + "grad_norm": 0.0001997270155698061, + "kept_token_count": 572.8984375, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899981379508972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9069895137799904e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5500061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5088256835937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2999547839164734, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.12874110639095307, + "grad_norm": 0.00022544543026015162, + "kept_token_count": 572.903125, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903591454029084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.765981855394784e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2247375488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19810, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3005388855934143, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.1361645795404911, + "grad_norm": 0.00020270417735446244, + "kept_token_count": 572.2875, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898152709007263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.978432432631962e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.83644714355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2097991943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19820, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.29909670650959014, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.13548240140080453, + "grad_norm": 0.00027653732104226947, + "kept_token_count": 573.5421875, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895879030227661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0672516843187625e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9379943847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5869110107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19830, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.3008462429046631, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.13091608434915541, + "grad_norm": 0.00035480354563333094, + "kept_token_count": 571.871875, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896417081356048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.046232534165029e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7042510986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4781860351562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19840, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.300204935669899, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.12765831649303436, + "grad_norm": 0.00025062603526748717, + "kept_token_count": 572.6359375, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893632888793945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.15498114307411e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1735565185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5780914306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.298896336555481, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.128217875957489, + "grad_norm": 0.00024453268270008266, + "kept_token_count": 573.65625, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897761344909668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9937137262313624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9545501708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1079498291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3007042735815048, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.13034760132431983, + "grad_norm": 0.00021727450075559318, + "kept_token_count": 572.209375, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900712847709656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.878417737723794e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.63030395507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7323455810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19870, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2984447985887527, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.12978823110461235, + "grad_norm": 0.00030989578226581216, + "kept_token_count": 574.225, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892288982868195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.207481542835012e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.35015563964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7601379394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19880, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.30027494132518767, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.132071416079998, + "grad_norm": 0.00017618939455132931, + "kept_token_count": 572.44375, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902439713478088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.810960151895415e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5110656738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.13853149414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19890, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30010311007499696, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.12511046528816222, + "grad_norm": 0.00018730245938058943, + "kept_token_count": 572.4546875, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899115085601806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.940836068068165e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7604614257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6072204589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19900, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.29989067316055296, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.13839449360966682, + "grad_norm": 0.00020753884746227413, + "kept_token_count": 572.759375, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9881609201431274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.6246591955423356e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2598541259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.30919494628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.30089797675609586, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.13524608239531516, + "grad_norm": 0.0002592884993646294, + "kept_token_count": 571.84375, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989336359500885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.16550556110451e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0402770996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.34015197753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19920, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.300079682469368, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.13466664552688598, + "grad_norm": 0.0003031268424820155, + "kept_token_count": 572.721875, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989518666267395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0942927444120866e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4148712158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4140228271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19930, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.300274932384491, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.12530431002378464, + "grad_norm": 0.00019745834288187325, + "kept_token_count": 572.6375, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896238744258881, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.053185948578175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.7481750488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.93963012695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19940, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30080549120903016, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.13151175156235695, + "grad_norm": 0.0002481875999365002, + "kept_token_count": 571.940625, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904697954654693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.722759429365396e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3074096679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6563690185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19950, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.30000119805336, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.1355790451169014, + "grad_norm": 0.000227903772611171, + "kept_token_count": 572.7765625, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896944701671601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.025612179248128e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4846954345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2642761230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19960, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2996908277273178, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.12187267541885376, + "grad_norm": 0.0002997741976287216, + "kept_token_count": 573.0140625, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907198786735535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6250671655579936e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.69529113769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1154846191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19970, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2997067987918854, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.12735893949866295, + "grad_norm": 0.00023514499480370432, + "kept_token_count": 572.8578125, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889727354049682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3075441499240694e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3940704345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.67478637695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19980, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.3002627432346344, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.12374587878584861, + "grad_norm": 0.00027177296578884125, + "kept_token_count": 572.5125, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892971456050873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.180820360488724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.70545654296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5773223876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19990, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3004376173019409, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.13077812418341636, + "grad_norm": 0.0001991671451833099, + "kept_token_count": 572.2671875, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989963710308075, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.920441959053278e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.81483764648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.27528686523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20000, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.2992779850959778, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.11915845721960068, + "grad_norm": 0.0002521232527215034, + "kept_token_count": 573.346875, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893253982067108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1697884080349466e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.75873413085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8633758544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20010, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3001829326152802, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.12252349257469178, + "grad_norm": 0.00023613758094143122, + "kept_token_count": 572.55, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898360967636108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9702916546957566e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8802520751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6859588623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20020, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3008534610271454, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.1300024315714836, + "grad_norm": 0.00021233016741462052, + "kept_token_count": 572.021875, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897584736347198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.000616972916759e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.75795593261716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.30028076171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20030, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29986645877361295, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.12063878774642944, + "grad_norm": 0.00032247923081740737, + "kept_token_count": 572.8734375, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990048635005951, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.887273560394533e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0520843505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.91038818359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20040, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.299825918674469, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.12089181020855903, + "grad_norm": 0.0001785677159205079, + "kept_token_count": 573.0078125, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900993347167969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.867460873152595e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8677001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3432281494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20050, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.30030265748500823, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.12243709415197372, + "grad_norm": 0.00019468528626020998, + "kept_token_count": 572.4875, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895867109298706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.067710651725065e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.18173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.31085510253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20060, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.3004053145647049, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.12146087065339088, + "grad_norm": 0.00016369497461710125, + "kept_token_count": 572.3703125, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902011632919312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.827692889899481e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1853759765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9444244384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20070, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.29935083091259, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.12869273349642754, + "grad_norm": 0.00034644396509975195, + "kept_token_count": 573.1953125, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895759403705597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0719183380133474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4025177001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.92498779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20080, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.300549191236496, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.1259709045290947, + "grad_norm": 0.00024492747616022825, + "kept_token_count": 572.4390625, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903541803359985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.767925627471413e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.94249572753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5218475341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20090, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.3000100165605545, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.12328922227025033, + "grad_norm": 0.00023224180040415376, + "kept_token_count": 572.8453125, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900273442268371, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.895587469742168e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.79941711425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0164093017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20100, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.29953778684139254, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.12164639458060264, + "grad_norm": 0.00017906502762343735, + "kept_token_count": 573.16875, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894462287425995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1225922177545724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9418426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3692565917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29959788620471955, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.12588457763195038, + "grad_norm": 0.0002554402453824878, + "kept_token_count": 573.1859375, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896207869052887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.054397577419877e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3301300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2591979980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.2992683917284012, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.12706262692809106, + "grad_norm": 0.00022933949367143214, + "kept_token_count": 573.265625, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899388015270233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.930172897526063e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7329833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6710510253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20130, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.29958787858486174, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.13400683626532556, + "grad_norm": 0.00037589113344438374, + "kept_token_count": 573.096875, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897273421287537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.012774516013451e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7942352294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9737884521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20140, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3004838705062866, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.12323860451579094, + "grad_norm": 0.00021471790387295187, + "kept_token_count": 572.2625, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897135019302368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0181817894335836e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.67738647460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9842498779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20150, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.29901480972766875, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.12488323375582695, + "grad_norm": 0.00018773625197354704, + "kept_token_count": 573.4703125, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897322833538056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.010846132587176e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.69151611328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6080749511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20160, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.2998940110206604, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.1232779510319233, + "grad_norm": 0.0002461077820044011, + "kept_token_count": 573.1203125, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898328125476837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.971577534684911e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.687841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4998474121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20170, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.3003030300140381, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.11933415159583091, + "grad_norm": 0.00020661055168602616, + "kept_token_count": 572.484375, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898786962032318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.953658051614184e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6111663818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.832373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20180, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.29959845542907715, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.12573479786515235, + "grad_norm": 0.00026009377324953675, + "kept_token_count": 573.19375, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892204761505127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.210772131045815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7089447021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7100006103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20190, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2994256854057312, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.12549926117062568, + "grad_norm": 0.00025001956964842975, + "kept_token_count": 573.065625, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899774730205536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.915067536581773e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7521514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.41904907226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20200, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2990769296884537, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.1383369378745556, + "grad_norm": 0.00019773130770772696, + "kept_token_count": 573.478125, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889049470424652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3340251068002544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3109436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8930633544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20210, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29972695708274844, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.13413274213671683, + "grad_norm": 0.0002664109051693231, + "kept_token_count": 573.109375, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892437160015106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.201694573566783e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.24306030273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5120483398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20220, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.3003463238477707, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.12943940758705139, + "grad_norm": 0.0003410828940104693, + "kept_token_count": 572.4765625, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894544959068299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.119360055483412e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3896118164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4205627441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20230, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.2997545599937439, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.11898788884282112, + "grad_norm": 0.00029761690529994667, + "kept_token_count": 572.95625, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899884402751923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9107821794459594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.07105407714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68102111816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20240, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29978362619876864, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.123732940107584, + "grad_norm": 0.0002759210183285177, + "kept_token_count": 572.740625, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899670481681824, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.919142254744656e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.90709533691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.64357604980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20250, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.3020786434412003, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.13938318938016891, + "grad_norm": 0.00022118240303825587, + "kept_token_count": 571.03125, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902056515216827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.825934218184557e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9532012939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9830718994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20260, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30014827847480774, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.12050047144293785, + "grad_norm": 0.00022517218894790858, + "kept_token_count": 572.6984375, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899172902107238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.938577938242815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.112451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.85093383789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20270, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.3006870597600937, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.11865042820572853, + "grad_norm": 0.00025271420599892735, + "kept_token_count": 572.2046875, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897795021533966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.992401907453314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.98382873535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.67218017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20280, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.30036792159080505, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.11974154338240624, + "grad_norm": 0.00022379662550520152, + "kept_token_count": 572.4734375, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902841210365295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.795285192609299e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.178271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.81829528808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20290, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29955429434776304, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.1281689703464508, + "grad_norm": 0.00019216947839595377, + "kept_token_count": 573.16875, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894335448741913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.127544852963183e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2197967529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4949920654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20300, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.30038459300994874, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.13412587568163872, + "grad_norm": 0.0002477025263942778, + "kept_token_count": 572.5171875, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894791901111603, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1097061694017614e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.55927124023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0223052978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30062997043132783, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.1261284776031971, + "grad_norm": 0.0002028880116995424, + "kept_token_count": 572.3734375, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900805950164795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.874790691043018e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3647979736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.67804260253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3010083347558975, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.12413509339094161, + "grad_norm": 0.00021634432778228074, + "kept_token_count": 572.1328125, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904989719390869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.711358986038249e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.860107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3537536621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20330, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.29965803027153015, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.12314220070838929, + "grad_norm": 0.00021944493346381932, + "kept_token_count": 573.2328125, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900240480899811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.896874113706872e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.00213317871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9695343017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20340, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.30057106018066404, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.13222475126385688, + "grad_norm": 0.00018636365712154657, + "kept_token_count": 572.284375, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892041623592377, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.217148944007931e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4564666748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.74969787597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30023046731948855, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.12241259589791298, + "grad_norm": 0.0002265858929604292, + "kept_token_count": 572.571875, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904072225093842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.747198534256313e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.427978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4020172119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20360, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29987284243106843, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.1333690859377384, + "grad_norm": 0.00019056478049606085, + "kept_token_count": 572.5890625, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99025838971138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.805338965321425e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1638244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.79248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20370, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.29959881901741026, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.13746435791254044, + "grad_norm": 0.00018375067156739533, + "kept_token_count": 573.1265625, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897353231906891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.009654512628913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0661376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2635955810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20380, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2988934814929962, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.12109802365303039, + "grad_norm": 0.00019377034914214164, + "kept_token_count": 573.64375, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990224152803421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.818710538325831e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1056427001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6339569091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20390, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.30048746764659884, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.1288117602467537, + "grad_norm": 0.00023061515821609646, + "kept_token_count": 572.2953125, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898685455322266, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9576221024617555e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.03020629882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0705108642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20400, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.3002136290073395, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.12445354089140892, + "grad_norm": 0.0002445199934300035, + "kept_token_count": 572.575, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896390497684479, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.047266920679249e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8447021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.517626953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20410, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.3006684958934784, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.12867906093597412, + "grad_norm": 0.00025848159566521645, + "kept_token_count": 572.31875, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990062826871872, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.881731245201081e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.32602844238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.640673828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20420, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2993953049182892, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.1305850028991699, + "grad_norm": 0.0001852631103247404, + "kept_token_count": 573.196875, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907594382762909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.609614796005189e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4265502929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0299041748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20430, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.3000813513994217, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.13359322398900986, + "grad_norm": 0.00022118541528470814, + "kept_token_count": 572.625, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900123953819275, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9014301910356154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9418182373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1706115722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20440, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.30013634264469147, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.12199864760041237, + "grad_norm": 0.00017836337792687118, + "kept_token_count": 572.753125, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901646971702576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.841931866190862e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.24456176757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4723358154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20450, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2991874128580093, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.12386086136102677, + "grad_norm": 0.0001889182603918016, + "kept_token_count": 573.271875, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900530815124512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.885528567479923e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9199920654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8050048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20460, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.30084327459335325, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.12612417414784433, + "grad_norm": 0.00025309313787147403, + "kept_token_count": 571.8765625, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894879817962646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1062773743760775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3528533935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0883453369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20470, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.2999807268381119, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.1450548216700554, + "grad_norm": 0.00016017777670640498, + "kept_token_count": 572.784375, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904517948627471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.729791733348975e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2823852539062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2568115234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20480, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.29960368275642396, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.13185301199555396, + "grad_norm": 0.00022724660811945796, + "kept_token_count": 573.2203125, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894694983959198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1134965431410817e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.25335998535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.28216247558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20490, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.2993683457374573, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.12355719059705735, + "grad_norm": 0.00021971555543132126, + "kept_token_count": 573.190625, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897474646568298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.004919428552966e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5766662597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3457885742188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20500, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.3007840573787689, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.12824912890791892, + "grad_norm": 0.00027745700208470225, + "kept_token_count": 571.9296875, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896171569824219, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0558085311204194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5032501220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.45509338378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20510, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.29970631301403045, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.11986046358942985, + "grad_norm": 0.00025583175010979176, + "kept_token_count": 573.0140625, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905146360397339, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.70524048776133e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3971008300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7623687744141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29953182935714723, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.12640395835042, + "grad_norm": 0.00023330967815127224, + "kept_token_count": 573.09375, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897374331951141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0088382229441776e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7313262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.51793212890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20530, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30011088252067564, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.1290796771645546, + "grad_norm": 0.00023308103845920414, + "kept_token_count": 572.6703125, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894536554813385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1196880920324476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.69183044433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.99847717285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20540, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2994422733783722, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.11588127464056015, + "grad_norm": 0.00021481029398273677, + "kept_token_count": 573.5515625, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902369379997253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.813705625361763e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1253936767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9501525878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20550, + "valid_token_count": 818.70625 + }, + { + "actual_token_mask_ratio": 0.29993458092212677, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.1267676331102848, + "grad_norm": 0.00021271493460517377, + "kept_token_count": 572.8125, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892264127731323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.208459686196875e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6524993896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4960021972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20560, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29904999434947965, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.12130558639764785, + "grad_norm": 0.00023485513520427048, + "kept_token_count": 573.6609375, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896528244018554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.041884058096912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2911041259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.11126403808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20570, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3001944363117218, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.12855651453137398, + "grad_norm": 0.0003774689685087651, + "kept_token_count": 572.528125, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895419895648956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.085178952664137e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.42510681152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7656982421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20580, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3001351714134216, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.1278749570250511, + "grad_norm": 0.00020639473223127425, + "kept_token_count": 572.434375, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908133387565613, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.58855784725165e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2380401611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8047668457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20590, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.29996259808540343, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.13271073251962662, + "grad_norm": 0.00017234269762411714, + "kept_token_count": 572.71875, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905518472194672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.690704070322681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.30413818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1593841552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20600, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.3001004844903946, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.13110419362783432, + "grad_norm": 0.00017593888333067298, + "kept_token_count": 572.790625, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895632266998291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.076881377841346e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.90430603027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.10230712890626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20610, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.3002393990755081, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.12502409294247627, + "grad_norm": 0.00027416920056566596, + "kept_token_count": 572.6875, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900638580322265, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8813288847450166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6096557617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6317169189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20620, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.3007763743400574, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.12706462517380715, + "grad_norm": 0.00027306482661515474, + "kept_token_count": 572.1109375, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894312858581543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1284224425908175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.29216918945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9780029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20630, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30089427828788756, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.13255908712744713, + "grad_norm": 0.00024413492064923048, + "kept_token_count": 571.9375, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899507462978363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9255045339814385e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.17299499511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.15281982421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.30019737780094147, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.11786730289459228, + "grad_norm": 0.00018701843509916216, + "kept_token_count": 572.5328125, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0, + "mgd_cos_sim": 0.98993901014328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.930092643713579e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.996484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8748046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20650, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2991582840681076, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.1315091885626316, + "grad_norm": 0.0002870334137696773, + "kept_token_count": 573.725, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892577946186065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1961934039136394e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7995178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75505981445315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20660, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.2994092971086502, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.12056682705879211, + "grad_norm": 0.0002488352474756539, + "kept_token_count": 573.5109375, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899181008338929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9382633258355784e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.10655212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3612945556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20670, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.30089148283004763, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.13395222648978233, + "grad_norm": 0.00019804493058472872, + "kept_token_count": 571.8140625, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896763145923615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.032709694001824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.48309936523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8553070068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20680, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.3005674988031387, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.12631420269608498, + "grad_norm": 0.0001944503019331023, + "kept_token_count": 572.5390625, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897054195404053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0213428292190656e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.29347839355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75535888671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20690, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.301284259557724, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.12332737147808075, + "grad_norm": 0.00018800649559125304, + "kept_token_count": 571.7765625, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9887401342391968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.3984024887322445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.58482055664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5118072509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.29947944581508634, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.12362970486283302, + "grad_norm": 0.00028468671371228993, + "kept_token_count": 573.25625, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896401584148407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0468337465426886e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.66440734863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9764770507812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20710, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.3005062729120255, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.1318303771317005, + "grad_norm": 0.0002117085095960647, + "kept_token_count": 572.43125, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899222135543824, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9366593773593196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3997467041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6150299072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20720, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.3001710742712021, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.1293075129389763, + "grad_norm": 0.00026639734278433025, + "kept_token_count": 572.7890625, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905128717422486, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.705930776050081e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.06407775878904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3939514160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20730, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.2987559765577316, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.12113719433546066, + "grad_norm": 0.0002234016137663275, + "kept_token_count": 573.8078125, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908800542354583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5624944939627315e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.397265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.70908203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20740, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3004355639219284, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.12313301488757133, + "grad_norm": 0.00024156599829439074, + "kept_token_count": 572.5953125, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99035085439682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.769213981286157e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9744171142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6373809814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20750, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.30009335577487944, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.12165402919054032, + "grad_norm": 0.00033762582461349666, + "kept_token_count": 572.6484375, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9890776872634888, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.266546529834159e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.20005493164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5141845703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20760, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3000671714544296, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.12931354865431785, + "grad_norm": 0.0002854717895388603, + "kept_token_count": 572.7890625, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902003824710846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8279989712464155e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.39976806640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.40089721679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20770, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2993414670228958, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.1254113033413887, + "grad_norm": 0.00031722933636046946, + "kept_token_count": 573.6109375, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900518655776978, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8860135464346965e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9016082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8789520263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20780, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.2996419757604599, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.12915567234158515, + "grad_norm": 0.00013951616710983217, + "kept_token_count": 572.9234375, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902700304985046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.800788472290151e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6357452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.20851745605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20790, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.2995551973581314, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.1183528371155262, + "grad_norm": 0.00032584407017566264, + "kept_token_count": 573.0171875, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902852058410645, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.794856420427095e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.69403076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8157531738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20800, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.299964040517807, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.1312149442732334, + "grad_norm": 0.0002468983002472669, + "kept_token_count": 572.9484375, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898607432842255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.96066068788059e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.29808044433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5789459228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20810, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2996091157197952, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.1205343596637249, + "grad_norm": 0.00021431299683172256, + "kept_token_count": 572.95, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898361325263977, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.97027706640074e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.17734985351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1205810546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20820, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.3002835214138031, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.13142159804701806, + "grad_norm": 0.0003430511278565973, + "kept_token_count": 572.6765625, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893747985363006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.150488857703749e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4640777587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.48180236816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2999265223741531, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.131496212631464, + "grad_norm": 0.0001890556450234726, + "kept_token_count": 572.8390625, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900389909744263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8910385410417805e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4772888183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.304443359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20840, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30012764036655426, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.1154661126434803, + "grad_norm": 0.00021831283811479807, + "kept_token_count": 572.6453125, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901281118392944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.856225303024985e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.04652404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.27199401855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20850, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2989853203296661, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.1334855154156685, + "grad_norm": 0.0002519063709769398, + "kept_token_count": 573.490625, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905559301376343, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.689109071274288e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9454071044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8436279296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20860, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.29982632398605347, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.1246802806854248, + "grad_norm": 0.00020303235214669257, + "kept_token_count": 572.9453125, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901736795902252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.838426528091077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1984497070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8918792724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20870, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2992733359336853, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.11924771890044213, + "grad_norm": 0.0001942753733601421, + "kept_token_count": 573.3046875, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908210158348083, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.585559497878421e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5806915283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2288421630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20880, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.301020485162735, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.13411681354045868, + "grad_norm": 0.00031168627901934087, + "kept_token_count": 572.0453125, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897204458713531, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.01546829380095e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3282928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7832092285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20890, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29904015064239503, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.13220496028661727, + "grad_norm": 0.00019509032426867634, + "kept_token_count": 573.78125, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899220824241638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.936701687052846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5858215332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.06650390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20900, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.2993330299854279, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.12069885805249214, + "grad_norm": 0.000263084628386423, + "kept_token_count": 573.34375, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896173119544983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.05575741751818e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.21942138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.74921264648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20910, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.29915631413459776, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.13994070142507553, + "grad_norm": 0.0002702752244658768, + "kept_token_count": 573.515625, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903546452522278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.767740854527801e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8876708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2974060058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20920, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2988869309425354, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.11486283242702484, + "grad_norm": 0.00020039781520608813, + "kept_token_count": 573.96875, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898841857910157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.951507460442372e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.93946228027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7523193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20930, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.3000215232372284, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.12314591705799102, + "grad_norm": 0.00024136219872161746, + "kept_token_count": 572.765625, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896693885326385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0354156226385385e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8431396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71355895996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20940, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29963832497596743, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.13272822722792627, + "grad_norm": 0.000259280001046136, + "kept_token_count": 572.959375, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902509808540344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8082293031038716e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.95982971191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.22276306152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30031355917453767, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.13587650209665297, + "grad_norm": 0.00021349487360566854, + "kept_token_count": 572.50625, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900591909885407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8831492202007214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3206817626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.48493041992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20960, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2988195776939392, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.1309904709458351, + "grad_norm": 0.0002260297624161467, + "kept_token_count": 573.8, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895631313323975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.076919503859244e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.42783508300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6802642822266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20970, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29859092831611633, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.13256748020648956, + "grad_norm": 0.00029201500001363456, + "kept_token_count": 574.115625, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898523151874542, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9639614260522646e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.33184509277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3557373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20980, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.29971139430999755, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.12002160847187042, + "grad_norm": 0.00022261080448515713, + "kept_token_count": 572.890625, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901505708694458, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8474515167763454e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.053662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8858581542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20990, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2999847561120987, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.11985848024487496, + "grad_norm": 0.00022333806555252522, + "kept_token_count": 572.7390625, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899965286254883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.907625468855258e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3304107666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8583160400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21000, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3004511117935181, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.13650819659233093, + "grad_norm": 0.0002653561532497406, + "kept_token_count": 572.446875, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904301583766937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.738241430255585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6135955810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2560089111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30014942288398744, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.1330537497997284, + "grad_norm": 0.00021382432896643877, + "kept_token_count": 572.440625, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893435299396515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1627059545135126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0488311767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3322448730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21020, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.301195952296257, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.12477061599493026, + "grad_norm": 0.00027491533546708524, + "kept_token_count": 571.9765625, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899495542049408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.925971432181541e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9122375488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0310760498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21030, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.29938114881515504, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.12985364124178886, + "grad_norm": 0.00026566936867311597, + "kept_token_count": 573.2671875, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898953914642334, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.947131590393837e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7185028076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.61126098632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21040, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30033983290195465, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.1292201556265354, + "grad_norm": 0.0003190681163687259, + "kept_token_count": 572.528125, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896597146987915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.039192281197756e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.21416625976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7149139404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21050, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30003978312015533, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.12085467875003815, + "grad_norm": 0.0002111979410983622, + "kept_token_count": 572.7453125, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990361088514328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7652139872079716e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5345489501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7149963378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21060, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29981075823307035, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.11842768341302871, + "grad_norm": 0.0002553814265411347, + "kept_token_count": 572.9625, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990394926071167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.751998738152906e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99555053710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.82379150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21070, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.29993894696235657, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.128929004073143, + "grad_norm": 0.0002908477617893368, + "kept_token_count": 572.928125, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9889791309833527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.305041547922883e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.16151428222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.16746520996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21080, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29922703504562376, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.12784768342971803, + "grad_norm": 0.0002706049708649516, + "kept_token_count": 573.4125, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902775406837463, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.797853169089649e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8748443603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63746032714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21090, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.3001547396183014, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.12940986603498458, + "grad_norm": 0.00024874135851860046, + "kept_token_count": 572.70625, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901797115802765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.836069990938995e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0494384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.59026489257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21100, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2991030991077423, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.13267057314515113, + "grad_norm": 0.0002020164392888546, + "kept_token_count": 573.415625, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902081966400147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8249398676271086e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.61375427246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5773040771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21110, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.3011484920978546, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.12235147207975387, + "grad_norm": 0.00025173582253046334, + "kept_token_count": 571.9640625, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892766833305359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.188815291854553e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.87489929199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.17701110839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21120, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2997537076473236, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.13408950567245484, + "grad_norm": 0.0002670132671482861, + "kept_token_count": 573.1, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989672964811325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.034016492369119e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8537322998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.92754821777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21130, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30021111071109774, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.12128313928842545, + "grad_norm": 0.0001935134787345305, + "kept_token_count": 572.415625, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899863839149475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.911591302312445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1871002197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6798065185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21140, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.29929872751235964, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.12918915674090387, + "grad_norm": 0.0002140973083442077, + "kept_token_count": 573.2265625, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903281152248382, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.778100617637392e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.61204833984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.48439025878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21150, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2999022305011749, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.1149238646030426, + "grad_norm": 0.00017497927183285356, + "kept_token_count": 572.73125, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901555359363556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8455118192359806e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8953491210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.61663208007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29926719665527346, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.12432326748967171, + "grad_norm": 0.00021471492073033005, + "kept_token_count": 573.6390625, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900733292102813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.877626331814099e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.10687866210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.99029235839845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21170, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.2996064782142639, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.12827470973134042, + "grad_norm": 0.00041798772872425616, + "kept_token_count": 573.2171875, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898374915122986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.969743702327833e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2366943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.35251770019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.29976743161678315, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.12492587342858315, + "grad_norm": 0.00029899238143116236, + "kept_token_count": 572.85, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896330058574676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.049631788802799e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9994354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2796325683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21190, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3009078651666641, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.11766127571463585, + "grad_norm": 0.00017571350326761603, + "kept_token_count": 572.190625, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899455726146698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9275286690099163e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.91173400878904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9118988037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21200, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.3012250423431396, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.12931919917464257, + "grad_norm": 0.00026997621171176434, + "kept_token_count": 571.6671875, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894264459609985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.130315101065207e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.81293334960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.777880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21210, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.2999792516231537, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.13335091397166252, + "grad_norm": 0.000175405788468197, + "kept_token_count": 572.8, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897746026515961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.994315666204784e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.55882568359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.28380126953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2999227553606033, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.1296851746737957, + "grad_norm": 0.0002998559211846441, + "kept_token_count": 572.646875, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899495601654053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.92596579331439e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7713928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0088226318359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21230, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.30067484080791473, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.13210313767194748, + "grad_norm": 0.00023420942306984216, + "kept_token_count": 572.3359375, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903900980949402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7538880496867934e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9305725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2086181640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21240, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.3006748050451279, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.1270833618938923, + "grad_norm": 0.0002454117056913674, + "kept_token_count": 572.2265625, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990050083398819, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.886707818310242e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7133331298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1700775146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21250, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.3001138687133789, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.12614012882113457, + "grad_norm": 0.0002527849283069372, + "kept_token_count": 572.634375, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894573450088501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.11824159527896e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2496704101562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.57265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21260, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3000764846801758, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.12801267504692077, + "grad_norm": 0.00024748346186243, + "kept_token_count": 572.553125, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0, + "mgd_cos_sim": 0.98999143242836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.909611477865837e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3283416748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3004974365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21270, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.300338140130043, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.12784442082047462, + "grad_norm": 0.00015639199409633875, + "kept_token_count": 572.43125, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905865907669067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.677130953292362e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.70107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.64130859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21280, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3011213421821594, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.13401361033320427, + "grad_norm": 0.00020927487639710307, + "kept_token_count": 571.9875, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900337159633636, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.893102111760527e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.74645080566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4639068603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21290, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2992225170135498, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.13023823052644729, + "grad_norm": 0.00020368536934256554, + "kept_token_count": 573.4625, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901131570339203, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.862067387672141e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4132019042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.65350036621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21300, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29903118014335633, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.13320852145552636, + "grad_norm": 0.0002505207958165556, + "kept_token_count": 573.6, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900978088378907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.868056883220561e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6907104492187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3529510498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21310, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3003363609313965, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.12266825139522552, + "grad_norm": 0.0002364832762395963, + "kept_token_count": 572.5265625, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900031924247742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.905027515429538e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4427825927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79193420410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21320, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.2984081894159317, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.12469344586133957, + "grad_norm": 0.00021718221250921488, + "kept_token_count": 574.159375, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903558015823364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.767282105400227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1929412841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8775604248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21330, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.30057657659053805, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.13699718043208123, + "grad_norm": 0.00021896135876886547, + "kept_token_count": 572.0921875, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897704780101776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9959291461855174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.62815246582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2476867675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21340, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.3004718810319901, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.13626206889748574, + "grad_norm": 0.00040614596218802035, + "kept_token_count": 572.3828125, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892495155334473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1994228740804826e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.33778991699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0177764892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2996262401342392, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.13352436944842339, + "grad_norm": 0.0002137852570740506, + "kept_token_count": 572.971875, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902139902114868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.822687103820499e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1223663330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.342041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21360, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.30004524290561674, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.1237711526453495, + "grad_norm": 0.00027208635583519936, + "kept_token_count": 572.6828125, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891532778739929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.23701923864428e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5746765136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7126953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21370, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.29869372844696046, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.1371457874774933, + "grad_norm": 0.0003097587323281914, + "kept_token_count": 573.8875, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895991623401642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.062844491272699e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.75792236328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65126647949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21380, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.29985466599464417, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.12860058918595313, + "grad_norm": 0.00018430779164191335, + "kept_token_count": 572.803125, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895740389823914, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0726618681219404e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7986297607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.299763435125351, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.13325679674744606, + "grad_norm": 0.0001764485496096313, + "kept_token_count": 572.921875, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9885493695735932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.47292401076993e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2162811279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9272491455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21400, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.30020600855350493, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.12467645108699799, + "grad_norm": 0.0002861906832549721, + "kept_token_count": 572.475, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895503759384155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.081899896846153e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7828643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7950714111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.299323695898056, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.13330865949392318, + "grad_norm": 0.00019552894809748977, + "kept_token_count": 573.1734375, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899287045001983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.934116612072103e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9599182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7610687255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21420, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.30049410164356233, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.12056693807244301, + "grad_norm": 0.00023760458861943334, + "kept_token_count": 572.428125, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898096799850464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9806180211598985e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.67046813964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2117156982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21430, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3004709392786026, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.1232808843255043, + "grad_norm": 0.0003089645761065185, + "kept_token_count": 572.359375, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901334583759308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.854139176837634e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.63294982910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0829620361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21440, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30040290355682375, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.12781099379062652, + "grad_norm": 0.00022629166778642684, + "kept_token_count": 572.56875, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902708172798157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.800480299105402e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.90721130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4357116699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21450, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.3006650507450104, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.1202661894261837, + "grad_norm": 0.00020674936240538955, + "kept_token_count": 572.2453125, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990532511472702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.698258260556031e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1022033691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.0976226806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21460, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.2996769815683365, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.12599198147654533, + "grad_norm": 0.00027944156317971647, + "kept_token_count": 573.1, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898651421070099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9589476000401194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.07252197265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.433056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21470, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.30045369565486907, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.1290341943502426, + "grad_norm": 0.00020837174088228494, + "kept_token_count": 572.3828125, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896336913108825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0493562119081615e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.00083923339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.72109069824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21480, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.2990850478410721, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.13233783543109895, + "grad_norm": 0.00021877806284464896, + "kept_token_count": 573.3453125, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904310941696167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7378739943960684e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1891662597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7424713134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21490, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.3002819508314133, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.11748329922556877, + "grad_norm": 0.00018534368427935988, + "kept_token_count": 572.5015625, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902031242847442, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8269229844445365e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.72511901855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8621765136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21500, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3005840539932251, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.12879102006554605, + "grad_norm": 0.0003024300094693899, + "kept_token_count": 572.3515625, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901714324951172, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8393031718442214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8051391601563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2368896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21510, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29881786108016967, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.13026761785149574, + "grad_norm": 0.0002844587143044919, + "kept_token_count": 573.8171875, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989782202243805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9913474029162896e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6718444824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9033447265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21520, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2994603395462036, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.12313157096505165, + "grad_norm": 0.00039105041651055217, + "kept_token_count": 573.440625, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905228078365326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.702054600580595e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.03955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9271942138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21530, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.2994285702705383, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.12613246068358422, + "grad_norm": 0.00045450954348780215, + "kept_token_count": 573.325, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989264291524887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.193657514406368e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.24329833984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.504638671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21540, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3004906803369522, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.13019564151763915, + "grad_norm": 0.0002163868775824085, + "kept_token_count": 572.30625, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896938025951385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.02587709686486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9952392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2812866210937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21550, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29972353875637053, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.1270408697426319, + "grad_norm": 0.00025586094125173986, + "kept_token_count": 573.153125, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990325528383255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.779106373258401e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8103546142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6160430908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21560, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.3002441465854645, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.12267163172364234, + "grad_norm": 0.000174615895957686, + "kept_token_count": 572.646875, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896641790866851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0374552190769465e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.57021179199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2190460205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21570, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2998264253139496, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.1302051194012165, + "grad_norm": 0.0001963547256309539, + "kept_token_count": 572.965625, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908432960510254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.576861636247486e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5652557373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1044525146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21580, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29885841012001035, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.12398132234811783, + "grad_norm": 0.0002324976958334446, + "kept_token_count": 573.7921875, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904744148254394, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.720952372532338e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.58726806640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3603271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21590, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.3015523493289948, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.1302180677652359, + "grad_norm": 0.00018108154472429305, + "kept_token_count": 571.6015625, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903944790363312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.752174343389925e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.93455505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.04082641601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21600, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.3005113750696182, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.12785207331180573, + "grad_norm": 0.00023503575357608497, + "kept_token_count": 572.11875, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896573543548584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.040118110424373e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9337982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.09471435546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21610, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.2998300284147263, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.11780943796038627, + "grad_norm": 0.00015819734835531563, + "kept_token_count": 573.009375, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990402352809906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.749100778804859e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.46222229003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3722686767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21620, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.30102678537368777, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.11553987264633178, + "grad_norm": 0.0003263422113377601, + "kept_token_count": 572.0921875, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900557637214661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.884480720444117e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2538726806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.19133911132815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21630, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.3011451900005341, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.12110616937279702, + "grad_norm": 0.00026055925991386175, + "kept_token_count": 571.703125, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902785897254944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7974461520207115e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.71502075195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.66380920410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21640, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.29909363090991975, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.12869076207280158, + "grad_norm": 0.0001979325315915048, + "kept_token_count": 573.7546875, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902207255363464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.820048877969384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5612854003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.13975830078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21650, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.30010451972484586, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.12756332978606225, + "grad_norm": 0.0002265317307319492, + "kept_token_count": 572.7328125, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99063640832901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6576753700501285e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.73108825683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.39484252929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21660, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2999163389205933, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.13296959400177003, + "grad_norm": 0.00026744662318378687, + "kept_token_count": 572.5515625, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990043830871582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8891522854100914e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6342498779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.90105590820315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21670, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.30087508261203766, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.13299673274159432, + "grad_norm": 0.0002260561304865405, + "kept_token_count": 572.15625, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896183431148529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.055354293086566e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.20911865234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7455078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21680, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2998933970928192, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.11703785136342049, + "grad_norm": 0.0002091637288685888, + "kept_token_count": 573.0015625, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901011347770691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8667699482175524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0492706298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.54864501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21690, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.3003142148256302, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.1333211325109005, + "grad_norm": 0.00019001643522642553, + "kept_token_count": 572.6265625, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896115064620972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.058028898725752e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1789245605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0194488525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21700, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.29906395077705383, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.13301782310009003, + "grad_norm": 0.0002102742000715807, + "kept_token_count": 573.5203125, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892890334129334, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.183989076409489e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.03406677246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3810241699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3001277923583984, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.1255545847117901, + "grad_norm": 0.0003209937713108957, + "kept_token_count": 572.728125, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908905982971191, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5583774479164274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.8276824951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.94809875488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21720, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.3001841366291046, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.12972632572054862, + "grad_norm": 0.00036467183963395655, + "kept_token_count": 572.390625, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898572087287902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.962045957450755e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3118072509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.41705017089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21730, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.2991816848516464, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.12501627951860428, + "grad_norm": 0.0001978865038836375, + "kept_token_count": 573.440625, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906320571899414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6593689219444056e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1783508300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21740, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.30066086649894713, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.12965262532234192, + "grad_norm": 0.00030872231582179666, + "kept_token_count": 572.0609375, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904890835285187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.71522346540587e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.43174438476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.30902404785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21750, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.30004640519618986, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.12397000789642335, + "grad_norm": 0.00020303778001107275, + "kept_token_count": 572.8421875, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900929868221283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.869947431667242e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8653839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3013916015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21760, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.300617977976799, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.13008732199668885, + "grad_norm": 0.00022943546355236322, + "kept_token_count": 572.28125, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906826138496398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6396252653503325e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1400665283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.193212890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21770, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2990555137395859, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.122388556599617, + "grad_norm": 0.0002445306454319507, + "kept_token_count": 573.5328125, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898889362812042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9496543831774036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0298675537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.56138000488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21780, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3003875404596329, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.11881539225578308, + "grad_norm": 0.00020674554980359972, + "kept_token_count": 572.5078125, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904387652873993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.734877536771819e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4156005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.25940856933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2997637212276459, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.12610735967755318, + "grad_norm": 0.0002510883496142924, + "kept_token_count": 572.8984375, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896496891975403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.043112603540067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8409057617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.90960693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21800, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2992631644010544, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.12371517047286033, + "grad_norm": 0.00018453098891768605, + "kept_token_count": 573.44375, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901817083358765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8352889532689005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0409332275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8332092285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30129557847976685, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.12924415916204451, + "grad_norm": 0.00026657056878320873, + "kept_token_count": 571.8484375, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900979936122895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.867987943522167e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8760192871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.15992736816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21820, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.2996111363172531, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.13063808605074884, + "grad_norm": 0.00024348212173208594, + "kept_token_count": 572.9375, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906086504459382, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6685178747575266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7657897949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2296447753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21830, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.299972802400589, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.1264839880168438, + "grad_norm": 0.0001968795550055802, + "kept_token_count": 572.65625, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902412056922912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.812042414210737e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.73569641113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.49280700683596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21840, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.30084610283374785, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.12919531315565108, + "grad_norm": 0.00018551675020717084, + "kept_token_count": 572.2234375, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907146155834198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6271222052164374e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5415985107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3634918212891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21850, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3001349329948425, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.12939309254288672, + "grad_norm": 0.0003377553657628596, + "kept_token_count": 572.596875, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990191912651062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.831303874903824e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.94887084960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0384552001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21860, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3005551129579544, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.12839618921279908, + "grad_norm": 0.00019296753453090787, + "kept_token_count": 572.4359375, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904196798801422, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.742331282410305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.14271240234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2594451904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21870, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.3010594189167023, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.12242263779044152, + "grad_norm": 0.00022109232668299228, + "kept_token_count": 571.9234375, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903592944145203, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.765916299016681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.10154113769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2744384765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29919424653053284, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.13468844145536424, + "grad_norm": 0.00029003649251535535, + "kept_token_count": 573.6859375, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904375076293945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7353676088969226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3744812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3485931396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21890, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.29925558865070345, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.12617354467511177, + "grad_norm": 0.00022094481391832232, + "kept_token_count": 573.4546875, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899962604045868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.907734462700318e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1538055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.73846435546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21900, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29922770261764525, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.12209103181958199, + "grad_norm": 0.00016724747547414154, + "kept_token_count": 573.425, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905446887016296, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.693499984365189e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6629364013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 347.64845581054686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21910, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29962013065814974, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.12468076646327972, + "grad_norm": 0.00039299781201407313, + "kept_token_count": 573.0796875, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901327908039093, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8544020571862345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.58592224121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7422790527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29928852021694186, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.13154854327440263, + "grad_norm": 0.00017604048480279744, + "kept_token_count": 573.459375, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898245692253113, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9747951450408434e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4479614257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4073120117188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21930, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.30016316175460817, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.12796780318021775, + "grad_norm": 0.0002378682402195409, + "kept_token_count": 572.7171875, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897440075874329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0062665721052324e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6785827636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3566192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21940, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30084334015846254, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.12889407351613044, + "grad_norm": 0.00021262778318487108, + "kept_token_count": 572.2875, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898252427577973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.974526334786788e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.37278137207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0711151123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21950, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29918632805347445, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.12076007947325706, + "grad_norm": 0.00018837119569070637, + "kept_token_count": 573.496875, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904336392879486, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.736882681550924e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.96959228515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1819763183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21960, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.30076744258403776, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.12674284130334854, + "grad_norm": 0.00020048899750690907, + "kept_token_count": 572.2328125, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906556904315948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.650136459327768e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7600372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4607788085938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21970, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.30021257400512696, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.13619177415966988, + "grad_norm": 0.00018864954472519457, + "kept_token_count": 572.5875, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896311283111572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.050358220410999e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5816375732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.74785766601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.30086368024349214, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.12111569344997405, + "grad_norm": 0.00019734480883926153, + "kept_token_count": 571.875, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898528099060059, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9637703230255285e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.98581848144534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21990, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2998077481985092, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.12774524092674255, + "grad_norm": 0.0002100733545375988, + "kept_token_count": 572.9, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901874184608459, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.833061182376696e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9538330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21167907714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22000, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3004796296358109, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.1280066639184952, + "grad_norm": 0.00025659488164819777, + "kept_token_count": 572.40625, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896306931972504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.050531024404336e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2427642822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9503234863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22010, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3002960056066513, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.12387578412890435, + "grad_norm": 0.0002541150024626404, + "kept_token_count": 572.546875, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908298432826996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5821072015096436e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0891510009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22020, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30058374106884, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.12757086232304574, + "grad_norm": 0.00026752299163490534, + "kept_token_count": 572.1734375, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907617747783661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608699844335206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3623901367188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.21092834472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22030, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.30053111016750333, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.12245108038187028, + "grad_norm": 0.0002973280497826636, + "kept_token_count": 572.3359375, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899690866470336, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.918343209079467e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.682470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6394348144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30048816800117495, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.12911117523908616, + "grad_norm": 0.0002411366003798321, + "kept_token_count": 572.3765625, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904735028743744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.721304165082984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2940948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5268890380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.29927538335323334, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.11896488294005395, + "grad_norm": 0.00022008907399140298, + "kept_token_count": 573.240625, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903851807117462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.755807229026687e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.17254028320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2821319580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22060, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30024667382240294, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.12479363903403282, + "grad_norm": 0.0002524422015994787, + "kept_token_count": 572.409375, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900760889053345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8765492718084715e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5346740722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77272644042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22070, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.30035272240638733, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.11703076660633087, + "grad_norm": 0.0001798361772671342, + "kept_token_count": 572.6109375, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898452818393707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9667042801738714e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.13869323730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75336303710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22080, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.300495108962059, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.12176237478852273, + "grad_norm": 0.0003431472578085959, + "kept_token_count": 572.3828125, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905016124248505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7103257272974585e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9632141113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.02280883789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22090, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2997137248516083, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.11763784065842628, + "grad_norm": 0.0002429514133837074, + "kept_token_count": 573.1875, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899569153785706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.923100593965501e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.41705017089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8564849853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22100, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.30047989785671236, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.13282510563731192, + "grad_norm": 0.00021873071091249585, + "kept_token_count": 572.61875, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899439930915832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.928145451936871e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4851776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5699981689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22110, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.30080402493476865, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.12778877690434456, + "grad_norm": 0.00024839420802891254, + "kept_token_count": 572.1296875, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899635434150695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.920509989256971e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.45807495117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1406219482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22120, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29990613758563994, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.1281200222671032, + "grad_norm": 0.0002997855481225997, + "kept_token_count": 573.203125, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901795268058777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.836145551758818e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5499694824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8520263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22130, + "valid_token_count": 818.75625 + }, + { + "actual_token_mask_ratio": 0.3003394573926926, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.12336724177002907, + "grad_norm": 0.0001874057052191347, + "kept_token_count": 572.4234375, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902277112007141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8173161738086495e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2821075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7825653076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22140, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.30002728998661043, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.12493575662374497, + "grad_norm": 0.00023366593813989311, + "kept_token_count": 572.5890625, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900231420993805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.897230562870391e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4578002929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.64528198242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22150, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.29938267469406127, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.13357471078634262, + "grad_norm": 0.000166632016771473, + "kept_token_count": 573.4109375, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897802710533142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9921033021528274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7576873779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4967498779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22160, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2996412068605423, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.1247666209936142, + "grad_norm": 0.0002206650096923113, + "kept_token_count": 573.090625, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906916737556457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6360810918267815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0279266357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.89088745117186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22170, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3007683604955673, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.13578338250517846, + "grad_norm": 0.00019577537022996694, + "kept_token_count": 572.00625, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905727207660675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.682545357150957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2114593505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5907897949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22180, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.3000431627035141, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.12424650192260742, + "grad_norm": 0.0002471564803272486, + "kept_token_count": 572.6125, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0, + "mgd_cos_sim": 0.98944451212883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.123256076127291e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1423004150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9005615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.29913710951805117, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.12483705803751946, + "grad_norm": 0.0002152194210793823, + "kept_token_count": 573.440625, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901610553264618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.843352860712912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4930816650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22200, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2996826320886612, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.13805069401860237, + "grad_norm": 0.0002130266511812806, + "kept_token_count": 573.078125, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897601068019867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.999980071967002e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2592437744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8835845947266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22210, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2991552323102951, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.11750714778900147, + "grad_norm": 0.0001957409258466214, + "kept_token_count": 573.4796875, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907133758068085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.627603364293463e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.96445617675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.48809814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22220, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3002690881490707, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.1228522963821888, + "grad_norm": 0.00019666958542075008, + "kept_token_count": 572.4171875, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908959448337555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5562913762987594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.51389465332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5657684326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.2998854070901871, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.12048293873667718, + "grad_norm": 0.00020878424402326345, + "kept_token_count": 572.9859375, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898285567760468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9732438381179236e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.10658569335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.41103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22240, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.2998399674892426, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.13215538933873178, + "grad_norm": 0.0002132812369382009, + "kept_token_count": 572.80625, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911272943019866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4659224002098196e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0762512207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.12933044433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22250, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2995129138231277, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.1181663729250431, + "grad_norm": 0.00035620160633698106, + "kept_token_count": 573.2859375, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905748963356018, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.681698581203818e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5054473876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7847381591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22260, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.30054965019226076, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.1305497407913208, + "grad_norm": 0.0002332573785679415, + "kept_token_count": 572.2515625, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898818612098694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.952417646360118e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.73241271972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6480651855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22270, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30125952064990996, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.13436237946152688, + "grad_norm": 0.00030488832271657884, + "kept_token_count": 571.5375, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894788384437561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1098485235124825e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8630645751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.497314453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22280, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.3001613825559616, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.12020886912941933, + "grad_norm": 0.00020231174130458385, + "kept_token_count": 572.7140625, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899411380290986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.929264566977508e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.21185302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.60048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22290, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2996951699256897, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.12878654077649115, + "grad_norm": 0.0003141176130156964, + "kept_token_count": 572.9484375, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897460281848908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0054810597212054e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4528747558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8707000732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22300, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.29924336373806, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.12217963337898255, + "grad_norm": 0.0002138970303349197, + "kept_token_count": 573.4546875, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901369273662567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.852785084745847e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.92714233398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8890899658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30031699538230894, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.12708256542682647, + "grad_norm": 0.00025669834576547146, + "kept_token_count": 572.4625, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897546470165253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.002114583272487e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.29876098632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.21051025390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22320, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.30005907118320463, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.1302497200667858, + "grad_norm": 0.00020392696023918688, + "kept_token_count": 572.921875, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895271420478821, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0909801100497134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1041564941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2347015380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22330, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.3002954453229904, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.11977236270904541, + "grad_norm": 0.00019180338131263852, + "kept_token_count": 572.2609375, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905967712402344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.673161954793613e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8171142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.42808227539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22340, + "valid_token_count": 817.8546875 + }, + { + "actual_token_mask_ratio": 0.30028204023838045, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.12757277116179466, + "grad_norm": 0.00020546330779325217, + "kept_token_count": 572.6703125, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990208500623703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.824821542366408e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0485076904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2934967041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30017594397068026, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.13201743736863136, + "grad_norm": 0.00021459579875227064, + "kept_token_count": 572.54375, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905614614486694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.686951095005497e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2160369873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7237915039062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22360, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30039010643959047, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.13206253796815873, + "grad_norm": 0.00018846198508981615, + "kept_token_count": 572.346875, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900739192962646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.877395138260908e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4380187988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.58387145996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22370, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30099116563796996, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.1256535194814205, + "grad_norm": 0.00024470913922414184, + "kept_token_count": 571.765625, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891458630561829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.239913396304473e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4412384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.474658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22380, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.3003769814968109, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.13353046253323556, + "grad_norm": 0.00022335865651257336, + "kept_token_count": 572.784375, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908786892890931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.563031968951691e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.04609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.15965576171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22390, + "valid_token_count": 818.7109375 + }, + { + "actual_token_mask_ratio": 0.3002522051334381, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.12367135509848595, + "grad_norm": 0.00027245315141044557, + "kept_token_count": 572.7921875, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895157873630523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.095414205949055e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4654083251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.50632629394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22400, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.29964637756347656, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.12340721860527992, + "grad_norm": 0.0002487923193257302, + "kept_token_count": 572.803125, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905442833900452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6936558899469674e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.11144104003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8353973388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22410, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.29975688457489014, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.14062508791685105, + "grad_norm": 0.00021045448374934494, + "kept_token_count": 573.28125, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904812335968017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.718289917742368e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3446350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.34874572753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22420, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.2998871147632599, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.12205894961953163, + "grad_norm": 0.0002701587218325585, + "kept_token_count": 573.046875, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990085631608963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.87281794246519e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.97130432128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.39308471679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22430, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2996831923723221, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.12130183428525924, + "grad_norm": 0.0002454213099554181, + "kept_token_count": 572.990625, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907807290554047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.601294411055278e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7085021972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.25787963867185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22440, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.30010764598846434, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.1265297330915928, + "grad_norm": 0.0001773487456375733, + "kept_token_count": 572.634375, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903485834598541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.77010292140767e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.66322326660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.91542053222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22450, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2995377004146576, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.12631484493613243, + "grad_norm": 0.00017901223327498883, + "kept_token_count": 573.184375, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910474479198456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.497112011245918e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2028076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63221435546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29992394149303436, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.13205456212162972, + "grad_norm": 0.00022635147615801543, + "kept_token_count": 572.75625, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990300577878952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.788861158682266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5936279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0058532714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22470, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3008338361978531, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.1253241054713726, + "grad_norm": 0.00021227440447546542, + "kept_token_count": 572.09375, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902882218360901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.793682080868166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.88226928710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2304351806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22480, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29976021945476533, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.1243519589304924, + "grad_norm": 0.0002284106012666598, + "kept_token_count": 573.0203125, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903326809406281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7763190630357715e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.23756713867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1749572753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22490, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.3002381831407547, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.12829517126083373, + "grad_norm": 0.00021704721439164132, + "kept_token_count": 572.5703125, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906388580799103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6567083952832036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.86280822753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2794158935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22500, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.2987842321395874, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.13206510990858078, + "grad_norm": 0.00023201899603009224, + "kept_token_count": 573.65, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898988842964173, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9457697130274025e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4284423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2403533935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22510, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2989731550216675, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.12973124384880066, + "grad_norm": 0.0001889229315565899, + "kept_token_count": 573.6375, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900273263454438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.895594018104021e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.80401306152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.52919006347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22520, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29925406277179717, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.1319122552871704, + "grad_norm": 0.00024097313871607184, + "kept_token_count": 573.4015625, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898636281490326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.959535933972802e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2670135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8830261230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3010974168777466, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.13602143675088882, + "grad_norm": 0.0002788764832075685, + "kept_token_count": 571.571875, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905115902423859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.706427232827991e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.48138427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6693939208984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22540, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.3000908136367798, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.12303555831313133, + "grad_norm": 0.0001945366821018979, + "kept_token_count": 572.4125, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900158286094666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9000886499707124e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3595855712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.62078247070315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22550, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.3004794180393219, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.12624230086803437, + "grad_norm": 0.00018957721476908773, + "kept_token_count": 572.371875, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99071706533432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.626164834713563e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.12281799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9903106689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22560, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.299069219827652, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.12304045930504799, + "grad_norm": 0.00022547670232597739, + "kept_token_count": 573.48125, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898098468780517, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9805557389627214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.60899047851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.83733825683595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2995263397693634, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.12774297073483468, + "grad_norm": 0.00020348884572740644, + "kept_token_count": 573.34375, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989503014087677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.100401420146227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.957568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.52709045410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22580, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.2999630272388458, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.12081224992871284, + "grad_norm": 0.00029059461667202413, + "kept_token_count": 572.6921875, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902599871158599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.804719490290154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.41180114746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0584350585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22590, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29939856827259065, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.1280415877699852, + "grad_norm": 0.0002638924343045801, + "kept_token_count": 573.225, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902262270450592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.817897777480539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.67591857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9160400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22600, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.299920117855072, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.12746643722057344, + "grad_norm": 0.0002012527547776699, + "kept_token_count": 573.040625, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892312109470367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.206576813885477e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9659423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5149200439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22610, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.2984248697757721, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.1279666543006897, + "grad_norm": 0.00032080619712360203, + "kept_token_count": 573.965625, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896875202655793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0283360067405735e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2780029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.975732421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22620, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.30149654150009153, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.13991031870245935, + "grad_norm": 0.00027796471840701997, + "kept_token_count": 571.7234375, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9892467439174653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.200505354674533e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.86240844726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9519836425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22630, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.2989105373620987, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.11934622004628181, + "grad_norm": 0.00026955659268423915, + "kept_token_count": 573.559375, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895638287067413, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.076642471773084e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.95338439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1513153076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22640, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.2997237741947174, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.12711310759186745, + "grad_norm": 0.00023532372142653912, + "kept_token_count": 572.7125, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898227691650391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.975491290475475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.13335876464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3081756591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22650, + "valid_token_count": 817.83125 + }, + { + "actual_token_mask_ratio": 0.30133571326732633, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.130125692486763, + "grad_norm": 0.0002327862021047622, + "kept_token_count": 571.75, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990373021364212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.760558029171079e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.250732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2470947265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22660, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.30024702548980714, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.1292678289115429, + "grad_norm": 0.00021697257761843503, + "kept_token_count": 572.6453125, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893385946750641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.164637721260078e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6688629150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.14710998535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22670, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2996177762746811, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.1255418874323368, + "grad_norm": 0.00029020648798905313, + "kept_token_count": 573.1359375, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898386716842651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9692905556876215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.92235412597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9162353515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22680, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30011670887470243, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.13463521152734756, + "grad_norm": 0.0002523901639506221, + "kept_token_count": 572.7578125, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900589525699616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.883239842252806e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1543792724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8518463134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22690, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2993356764316559, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.12298266962170601, + "grad_norm": 0.0001578441879246384, + "kept_token_count": 573.19375, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990041172504425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.890187399520073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.48357543945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.44481506347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22700, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.29963043332099915, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.1285034276545048, + "grad_norm": 0.00022886868100613356, + "kept_token_count": 572.9515625, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897954225540161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.986179617641028e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.95842895507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6035919189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22710, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.29997965395450593, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.12498716562986374, + "grad_norm": 0.0003588949330151081, + "kept_token_count": 572.9359375, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989496773481369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1028424675459974e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1174835205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3300384521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22720, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.3000780284404755, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.1277560219168663, + "grad_norm": 0.00018802208069246262, + "kept_token_count": 572.725, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895744860172272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.072486408404075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7928192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4799865722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22730, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2998595654964447, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.12827179953455925, + "grad_norm": 0.00023725666687823832, + "kept_token_count": 572.6796875, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910887956619263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.480957147985464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2194488525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8656005859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22740, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.3011359959840775, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.12431047037243843, + "grad_norm": 0.0002987396437674761, + "kept_token_count": 571.9390625, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898047506809234, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9825438580010084e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8380615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7173156738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22750, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30037363767623904, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.12988505959510804, + "grad_norm": 0.00021141978504601866, + "kept_token_count": 572.384375, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903772056102753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7589196290355174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.35364685058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.82818908691405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22760, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2998016655445099, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.11941934227943421, + "grad_norm": 0.0002083402796415612, + "kept_token_count": 572.9, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904758989810943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.72036960470723e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.07317810058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.57346496582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.299942335486412, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.13136054947972298, + "grad_norm": 0.0002926844754256308, + "kept_token_count": 572.8484375, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895252764225007, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.091708633495728e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1710632324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2093475341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29954244792461393, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.1284575119614601, + "grad_norm": 0.00016235921066254377, + "kept_token_count": 572.99375, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910583794116974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.492843115964206e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8685729980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9590789794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22790, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29951025247573854, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.11319727376103401, + "grad_norm": 0.00019278901163488626, + "kept_token_count": 573.3640625, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99095498919487, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.533230628818274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7515106201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9494964599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22800, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.29949943721294403, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.12773140296339988, + "grad_norm": 0.00027321657398715615, + "kept_token_count": 573.13125, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904559135437012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7281817276380026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4629791259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2059844970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22810, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2990611732006073, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.11801437810063362, + "grad_norm": 0.00025806797202676535, + "kept_token_count": 573.584375, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904445469379425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.732619734364562e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7015106201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.322705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22820, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.29970415532588957, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.11633297950029373, + "grad_norm": 0.00020132756617385894, + "kept_token_count": 573.3296875, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906499803066253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6523716335068454e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.817529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.82793579101565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22830, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.298785600066185, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.13207115456461907, + "grad_norm": 0.00029660522704944015, + "kept_token_count": 573.584375, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895701944828034, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.074159260198939e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.47008056640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5530700683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22840, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.29989500939846037, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.1190961942076683, + "grad_norm": 0.00020803116785828024, + "kept_token_count": 572.934375, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899795889854431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9142387686297296e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.78700866699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5433807373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2998497307300568, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.1211999237537384, + "grad_norm": 0.00017529685283079743, + "kept_token_count": 573.040625, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902118742465973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.823502593149897e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.35076599121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8620330810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22860, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.29919893741607667, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.14144764170050622, + "grad_norm": 0.00016682145360391587, + "kept_token_count": 573.165625, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896385729312897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0474555135006085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.27503967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4110931396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22870, + "valid_token_count": 817.8671875 + }, + { + "actual_token_mask_ratio": 0.30017369389534, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.1281123101711273, + "grad_norm": 0.00018141177133657038, + "kept_token_count": 572.59375, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990405160188675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.748005401575938e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8893981933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.52692260742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22880, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.30105116665363313, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.13324827775359155, + "grad_norm": 0.00023899679945316166, + "kept_token_count": 571.7296875, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903008878231049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7887310463702305e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0500091552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22890, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.30039357841014863, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.12925979122519493, + "grad_norm": 0.0002172053063986823, + "kept_token_count": 572.4453125, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902197003364563, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.820451638603117e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4699737548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6071533203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22900, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30064319670200346, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.12051781117916108, + "grad_norm": 0.0002757793408818543, + "kept_token_count": 572.2328125, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907802879810333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.601467305998085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9123901367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.36305847167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30059581995010376, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.139148710668087, + "grad_norm": 0.0002913794887717813, + "kept_token_count": 572.1375, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902804911136627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7967003663652577e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2143310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.91803588867185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22920, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.30002056062221527, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.13045388981699943, + "grad_norm": 0.0001915485627250746, + "kept_token_count": 572.80625, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900554120540619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8846266033942814e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.62977600097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.97084350585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22930, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.2992394596338272, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.13118710592389107, + "grad_norm": 0.0003079353482462466, + "kept_token_count": 573.384375, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901483476161956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.848319247481413e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7842529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9706329345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22940, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.3002193450927734, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.1300601080060005, + "grad_norm": 0.00023327032977249473, + "kept_token_count": 572.6140625, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905494272708892, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.691643469210249e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9596710205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.74587097167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3005900174379349, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.12634048238396645, + "grad_norm": 0.00023556531232316047, + "kept_token_count": 572.2359375, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904681146144867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.723416084540076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9894714355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7830291748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22960, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30001274347305296, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.13248216286301612, + "grad_norm": 0.0003168762195855379, + "kept_token_count": 572.834375, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899020195007324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.944548370782286e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5568115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7129425048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22970, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2994306981563568, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.1312984399497509, + "grad_norm": 0.0001956680789589882, + "kept_token_count": 573.76875, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899331629276276, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9323803866864183e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.16806945800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9103149414062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22980, + "valid_token_count": 819.003125 + }, + { + "actual_token_mask_ratio": 0.3007684826850891, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.13590119183063507, + "grad_norm": 0.00025467557134106755, + "kept_token_count": 572.253125, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898068964481354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.981698064308148e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.6648315429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.56415405273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22990, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3010833919048309, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.12303686738014222, + "grad_norm": 0.00018078101857099682, + "kept_token_count": 571.8546875, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989949208498001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9261098572751506e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5213897705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2270568847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23000, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3005135327577591, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.1254739545285702, + "grad_norm": 0.0003286659193690866, + "kept_token_count": 572.2703125, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899019300937653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9445738002541476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.40979919433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2881317138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23010, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30105711817741393, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.12261466681957245, + "grad_norm": 0.00022876268485561013, + "kept_token_count": 571.884375, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901764392852783, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.837347539956681e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.97080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1459045410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23020, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29916419088840485, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.12840562388300897, + "grad_norm": 0.00021225723321549594, + "kept_token_count": 573.3765625, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990092134475708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.870281507261098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9035278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7843475341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23030, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.2997158318758011, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.12609481438994408, + "grad_norm": 0.00024204912187997252, + "kept_token_count": 573.0421875, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903698801994324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.76178439182695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.66171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3369171142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23040, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30067830085754393, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.14131302013993263, + "grad_norm": 0.00027479027630761266, + "kept_token_count": 572.2296875, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898118376731873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.979771609010641e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9572418212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.41221008300784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23050, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2994768738746643, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.12618405967950821, + "grad_norm": 0.00021885927708353847, + "kept_token_count": 573.34375, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900972068309783, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.868297426379286e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.05128173828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.64404296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23060, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.29928758144378664, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.12011167705059052, + "grad_norm": 0.00025156073388643563, + "kept_token_count": 573.5859375, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896251142024994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.052708136441652e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2812927246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.35805053710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23070, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.3002834677696228, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.1325581707060337, + "grad_norm": 0.00026845739921554923, + "kept_token_count": 572.6625, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894518196582794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1204031003871935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5058624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.83954467773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23080, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.30024646520614623, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.12216485440731048, + "grad_norm": 0.00017237804422620684, + "kept_token_count": 572.5796875, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900950789451599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8691287954861764e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.51368408203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.86865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23090, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30038642287254336, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.13095131665468215, + "grad_norm": 0.00025759157142601907, + "kept_token_count": 572.571875, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99058096408844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.67933160305256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.20147705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6866760253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23100, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29953319430351255, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.12600954845547677, + "grad_norm": 0.0002065834414679557, + "kept_token_count": 573.3109375, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904425978660584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.733377325261244e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8536010742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.4484893798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23110, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.29996033012866974, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.12724151089787483, + "grad_norm": 0.00019620625243987888, + "kept_token_count": 572.8359375, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904354155063629, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.736180497071473e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3719085693359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2166809082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23120, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3008022904396057, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.12457484751939774, + "grad_norm": 0.0002305921516381204, + "kept_token_count": 572.234375, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906993210315704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.633098367572529e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.0985961914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.0352081298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23130, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29997746348381044, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.12489679232239723, + "grad_norm": 0.0002895224606618285, + "kept_token_count": 572.9796875, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898622155189514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9600887248525396e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0756011962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1012298583984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23140, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2996833175420761, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.12646240293979644, + "grad_norm": 0.00021841941634193063, + "kept_token_count": 573.18125, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990625923871994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.661769405880477e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.930712890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3004505455493927, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.1362512730062008, + "grad_norm": 0.00021862938592676073, + "kept_token_count": 572.4515625, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898548901081086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.962952941947151e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7702972412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.41831970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23160, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.30002312660217284, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.12343993708491326, + "grad_norm": 0.000300692132441327, + "kept_token_count": 572.4828125, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904476463794708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7314100518415215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2267181396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.30419006347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23170, + "valid_token_count": 817.85625 + }, + { + "actual_token_mask_ratio": 0.299998340010643, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.12358266040682793, + "grad_norm": 0.00021448393817991018, + "kept_token_count": 572.9, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904007971286773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7497092853300275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.52197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.79905700683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23180, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29999538958072663, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.1290186747908592, + "grad_norm": 0.0002204078482463956, + "kept_token_count": 573.040625, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901114046573639, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8627570029348134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5543914794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1033142089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23190, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.3005202919244766, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.12054511681199073, + "grad_norm": 0.0002614967816043645, + "kept_token_count": 572.5, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900665998458862, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.880254189425614e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.08338317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.265966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3004222810268402, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.12473160102963447, + "grad_norm": 0.00022953339794185013, + "kept_token_count": 572.6796875, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905440330505371, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.693754370033275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.49473571777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9294006347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23210, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.30066781044006347, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.13071116283535958, + "grad_norm": 0.00023975252406671643, + "kept_token_count": 572.2453125, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905672311782837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.684688181238016e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0826721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43280334472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23220, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.3002416640520096, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.1222816064953804, + "grad_norm": 0.0002212257677456364, + "kept_token_count": 572.4578125, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903163731098175, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7826809420948845e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9674011230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2630981445312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23230, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.29964863657951357, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.12879112362861633, + "grad_norm": 0.00022133052698336542, + "kept_token_count": 573.221875, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901117146015167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.862638150167186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6114837646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.05201110839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23240, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.29947017431259154, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.1332497552037239, + "grad_norm": 0.0001828540553105995, + "kept_token_count": 573.1625, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9891716003417969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.229859587212559e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.78538513183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3011474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23250, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2997465878725052, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.1374614305794239, + "grad_norm": 0.00019184511620551348, + "kept_token_count": 572.9109375, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899226903915406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.936468237952795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2605865478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.44631042480466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23260, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3005252152681351, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.13037632331252097, + "grad_norm": 0.00019920313206966966, + "kept_token_count": 572.34375, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902742564678192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.799138394242618e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3778503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7000335693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23270, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29917536675930023, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.12656922191381453, + "grad_norm": 0.00017698464216664433, + "kept_token_count": 573.390625, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897189497947693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.016055281681474e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7218505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5545654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.3001305162906647, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.13256509155035018, + "grad_norm": 0.00019829098891932517, + "kept_token_count": 572.85625, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898932814598084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9479614861193114e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9499908447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9614227294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23290, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.29942736625671384, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.11730158478021621, + "grad_norm": 0.00023634082754142582, + "kept_token_count": 573.284375, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906465232372283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.653722342278343e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8993316650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8830902099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23300, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30039042234420776, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.12515127509832383, + "grad_norm": 0.0001700141147011891, + "kept_token_count": 572.540625, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901553630828858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.845577957690693e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6078277587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.94794311523435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23310, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.30143978595733645, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.12285394221544266, + "grad_norm": 0.000234727151109837, + "kept_token_count": 571.73125, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906775712966919, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6415949762158564e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.26255493164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0207214355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23320, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.30009137392044066, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.126907616853714, + "grad_norm": 0.00018350528262089938, + "kept_token_count": 572.709375, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900408864021302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.890297266480047e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8903411865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3819915771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23330, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2997035801410675, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.13167161792516707, + "grad_norm": 0.00021912909869570285, + "kept_token_count": 572.9578125, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904574573040008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.727577641257085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6117248535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9353546142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23340, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3004154056310654, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.12979386150836944, + "grad_norm": 0.00026103309937752783, + "kept_token_count": 572.4109375, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900967121124268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.868490966851823e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8706848144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.98231506347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29920717775821687, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.1282213255763054, + "grad_norm": 0.00017891739844344556, + "kept_token_count": 573.234375, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990078866481781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.875459242408397e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.04607543945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.54515686035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23360, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.30084858238697054, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.1338910773396492, + "grad_norm": 0.00026071685715578496, + "kept_token_count": 572.0640625, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894736230373382, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.111884591111448e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.94512939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3026153564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23370, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.3003081113100052, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.1334440939128399, + "grad_norm": 0.00021340500097721815, + "kept_token_count": 572.5984375, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990830945968628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.581681667128578e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9806823730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7130828857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23380, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.3007627844810486, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.13343958854675292, + "grad_norm": 0.00023831702128518373, + "kept_token_count": 571.93125, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896710515022278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0347681715502406e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6852294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8971649169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23390, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.2997802674770355, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.13487956598401069, + "grad_norm": 0.00020319863688200712, + "kept_token_count": 572.990625, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906455338001251, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.654112915683072e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5775085449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1747680664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23400, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3008977949619293, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.12125025987625122, + "grad_norm": 0.0001653393410379067, + "kept_token_count": 572.0359375, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902416050434113, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.811891583609395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1075927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0567657470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.29986307621002195, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.12960526198148728, + "grad_norm": 0.00019156403141096234, + "kept_token_count": 573.0875, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906440496444702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.654687207017559e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.171142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2008422851562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23420, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.2998123824596405, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.1278804525732994, + "grad_norm": 0.00016616129141766578, + "kept_token_count": 572.796875, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903574168682099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.766651152545819e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3889190673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4576782226562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23430, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.3012694984674454, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.1266244664788246, + "grad_norm": 0.0002523119037505239, + "kept_token_count": 571.5203125, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906488239765168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.65281803169637e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.61509704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77366638183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23440, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30053777396678927, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.12142949551343918, + "grad_norm": 0.00023253470135387033, + "kept_token_count": 572.4609375, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900813043117523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8745106394344475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0719329833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3246612548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23450, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.2989886462688446, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.12494783997535705, + "grad_norm": 0.00023045370471663773, + "kept_token_count": 573.5078125, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901547491550445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8458197013824244e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5313659667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6435943603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23460, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.3009733438491821, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.12875365614891052, + "grad_norm": 0.00017995029338635504, + "kept_token_count": 571.853125, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990677934885025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6414514397620223e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3742156982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.73626708984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23470, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.29918256402015686, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.1357396624982357, + "grad_norm": 0.0001781816827133298, + "kept_token_count": 573.471875, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899904906749726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.909982624463737e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8501892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0704803466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3000036895275116, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.12581471726298332, + "grad_norm": 0.0002533669176045805, + "kept_token_count": 572.7765625, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899047791957856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9434648351743816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.28162841796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.45265808105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23490, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2995066732168198, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.1286211021244526, + "grad_norm": 0.00017206132179126143, + "kept_token_count": 573.0765625, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990478777885437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.719243377418024e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0895538330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0600952148437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23500, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.3002662628889084, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.12177745774388313, + "grad_norm": 0.00018715682381298393, + "kept_token_count": 572.521875, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909704029560089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5272001878183803e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.50404663085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5684509277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23510, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2996039420366287, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.13002340197563172, + "grad_norm": 0.00033143479959107935, + "kept_token_count": 573.2515625, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990390294790268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.753806922759395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1070983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4489685058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23520, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.29891441464424134, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.12527145966887474, + "grad_norm": 0.0001825451327022165, + "kept_token_count": 573.8453125, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902815639972686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7962790884193964e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.06004943847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.65997619628905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23530, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.30098690688610075, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.1322949171066284, + "grad_norm": 0.00016974365280475467, + "kept_token_count": 572.321875, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906944870948792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.63498882506974e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3251007080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.71266174316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23540, + "valid_token_count": 818.759375 + }, + { + "actual_token_mask_ratio": 0.3002349823713303, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.12932450100779533, + "grad_norm": 0.00019355428230483085, + "kept_token_count": 572.6421875, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903926610946655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7528897519223395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.17470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.88695373535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23550, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.29894145727157595, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.13124414458870887, + "grad_norm": 0.00023575915838591754, + "kept_token_count": 573.4359375, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902888536453247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7934382635285144e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8747802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0284729003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23560, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.29985928237438203, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.11954394876956939, + "grad_norm": 0.00030452682403847575, + "kept_token_count": 573.0953125, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990389484167099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.75412855646573e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.21970520019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.94033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23570, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.29928788244724275, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.12491478249430657, + "grad_norm": 0.00026465891278348863, + "kept_token_count": 573.471875, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904940664768219, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.713280893862248e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6631317138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.87030334472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23580, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29977748095989226, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.12691275626420975, + "grad_norm": 0.00021806563017889857, + "kept_token_count": 573.0390625, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899332523345947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9323414966929704e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1199920654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.156689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23590, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2995557188987732, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.12236026972532273, + "grad_norm": 0.00022601998352911323, + "kept_token_count": 573.18125, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990545243024826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.693290236697067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5121795654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.20912170410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23600, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3004806011915207, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.12179373130202294, + "grad_norm": 0.00024371362815145403, + "kept_token_count": 572.246875, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904455065727233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.732244549610186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7478485107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1775146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23610, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.2993248164653778, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.13275829404592515, + "grad_norm": 0.00021951076632831246, + "kept_token_count": 573.278125, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990574961900711, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6816723149968314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.83704528808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8563995361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23620, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.30051641166210175, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.12023202180862427, + "grad_norm": 0.0002858592488337308, + "kept_token_count": 572.5484375, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990860515832901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.570132648746949e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.98468627929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9846527099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23630, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.2995164692401886, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.12629228904843331, + "grad_norm": 0.0002231060789199546, + "kept_token_count": 573.175, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905544102191925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.689704644784797e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0482513427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.71812744140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29932954609394075, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.12540942952036857, + "grad_norm": 0.0002877057413570583, + "kept_token_count": 573.2828125, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904812455177308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7182846790528855e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1288665771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2450653076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23650, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.30116587579250337, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.1307595767080784, + "grad_norm": 0.00018872438522521406, + "kept_token_count": 571.7796875, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911585807800293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4536964994913434e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8127716064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0094421386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23660, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2990782052278519, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.1316199153661728, + "grad_norm": 0.0002046761364908889, + "kept_token_count": 573.6328125, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904851615428925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.71674817870371e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.11224365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4314666748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23670, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3005704373121262, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.12501016706228257, + "grad_norm": 0.000284048990579322, + "kept_token_count": 572.24375, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9895227193832398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.092712733836379e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4462036132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61376953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23680, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3004213094711304, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.12552285492420195, + "grad_norm": 0.0002055030345218256, + "kept_token_count": 572.5421875, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903670191764832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.762902670132462e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.07066955566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.57879028320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23690, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2991626799106598, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.13211918100714684, + "grad_norm": 0.0002689426764845848, + "kept_token_count": 573.315625, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990065598487854, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8806468728580513e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.36736755371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9255340576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23700, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.3006825715303421, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.12890729904174805, + "grad_norm": 0.0002247057855129242, + "kept_token_count": 572.065625, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908163666725158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.587380015233066e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.52586975097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7694061279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23710, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.2994716793298721, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.13358547389507294, + "grad_norm": 0.00020765159570146352, + "kept_token_count": 573.2203125, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900678098201752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.879781834257301e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.78037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7018707275391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23720, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29976127445697787, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.12750578969717025, + "grad_norm": 0.0002051909250440076, + "kept_token_count": 572.921875, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904218673706054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7414747203001754e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4633331298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5045227050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23730, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2997743487358093, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.12476571351289749, + "grad_norm": 0.00018605439981911331, + "kept_token_count": 573.0796875, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906683683395385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6451913183555006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.129736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.46026306152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23740, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.3002804905176163, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.12508225291967393, + "grad_norm": 0.00019944556697737426, + "kept_token_count": 572.746875, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900475203990936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.887705206579994e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1906768798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.11707763671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23750, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.30104399025440215, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.12842586934566497, + "grad_norm": 0.00019472504209261388, + "kept_token_count": 571.93125, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906439363956452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.654728898254689e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0539886474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7799835205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2999786972999573, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.1303366057574749, + "grad_norm": 0.00024471248616464436, + "kept_token_count": 572.771875, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897507548332214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.003632020612713e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7388946533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0945068359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23770, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.3000677675008774, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.13050443381071092, + "grad_norm": 0.00026107599842362106, + "kept_token_count": 572.7296875, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908690214157104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.566808536561439e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.04233703613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4053588867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23780, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30003832280635834, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.13627137020230293, + "grad_norm": 0.0001941847731359303, + "kept_token_count": 572.6578125, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901831388473511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.834730159724131e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5064697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0475341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23790, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2993173152208328, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.1276589274406433, + "grad_norm": 0.0001958964130608365, + "kept_token_count": 573.2, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903434813022614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.772095951717347e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.930029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.13219299316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23800, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.2996801406145096, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.13720299899578095, + "grad_norm": 0.00028259720420464873, + "kept_token_count": 573.159375, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900980293750763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.867975901812315e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7600616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3196746826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23810, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3007269948720932, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.13082418441772461, + "grad_norm": 0.0002519482804927975, + "kept_token_count": 572.109375, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989743995666504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0062647894956174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9822174072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23820, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2998370587825775, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.12746055349707602, + "grad_norm": 0.00023908847651910037, + "kept_token_count": 572.9390625, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902686953544617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8013089579180814e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7106414794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.78392028808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23830, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.30115973949432373, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.12116809040307999, + "grad_norm": 0.00019463698845356703, + "kept_token_count": 571.9765625, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989795035123825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.986325828009285e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0100799560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.08765563964846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23840, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.3005749940872192, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.13592843040823938, + "grad_norm": 0.00020423672685865313, + "kept_token_count": 572.3125, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898392140865326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.969071403844282e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7920684814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2247741699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23850, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.30016596615314484, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.12230894267559052, + "grad_norm": 0.00025691938935779035, + "kept_token_count": 572.55625, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903603196144104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7655217965948395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.87880249023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8090087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23860, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.29966171383857726, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.13657570034265518, + "grad_norm": 0.00018894614186137915, + "kept_token_count": 573.196875, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906100571155548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.667963574116584e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.39223022460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6238983154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23870, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2996388852596283, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.1170795887708664, + "grad_norm": 0.00016794812108855695, + "kept_token_count": 573.0953125, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990364158153534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.764023058465682e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.65577697753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.78327026367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23880, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3005042612552643, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.12402837350964546, + "grad_norm": 0.00023188983323052526, + "kept_token_count": 572.590625, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899877429008483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.911056410288438e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2654083251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.30078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23890, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.30107244849205017, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.14416610598564147, + "grad_norm": 0.00018857156101148576, + "kept_token_count": 571.9203125, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908413231372833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.57762184648891e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.44920959472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8417175292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23900, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29999385476112367, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.12200820818543434, + "grad_norm": 0.0001726109767332673, + "kept_token_count": 572.9984375, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9912133991718293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.432284211157821e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9457580566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9067810058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23910, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.3005214184522629, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.11984328478574753, + "grad_norm": 0.00022264569997787476, + "kept_token_count": 572.6140625, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904990911483764, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.711311182996724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5573272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8294403076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23920, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.299997478723526, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.1312417909502983, + "grad_norm": 0.0001954239996848628, + "kept_token_count": 572.859375, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900977969169616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.868062922265381e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.47040405273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7875518798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23930, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.29942179322242735, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.12863012850284578, + "grad_norm": 0.00022142735542729497, + "kept_token_count": 573.38125, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903151571750641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.783157844736706e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.87342224121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5500885009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23940, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.2996520012617111, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.13660307228565216, + "grad_norm": 0.000253400212386623, + "kept_token_count": 573.296875, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901077210903168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.86419334972743e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.40729064941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9961761474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23950, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.2998241066932678, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.13202443048357965, + "grad_norm": 0.00017320852202828974, + "kept_token_count": 573.171875, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990298581123352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.789632646657992e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7773834228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75376586914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23960, + "valid_token_count": 818.61875 + }, + { + "actual_token_mask_ratio": 0.3010797619819641, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.1287304386496544, + "grad_norm": 0.00024305465922225267, + "kept_token_count": 571.8359375, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899313628673554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.933083935407922e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7425231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.872998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2997866988182068, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.12281675413250923, + "grad_norm": 0.00017241948808077723, + "kept_token_count": 572.9421875, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906976997852326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6337331766844724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8259613037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.14837951660155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23980, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.3005522727966309, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.11729216501116753, + "grad_norm": 0.00020164439047221094, + "kept_token_count": 572.4265625, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903156399726868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.782970452448353e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0215667724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.65857849121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29970844089984894, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.1201419211924076, + "grad_norm": 0.000230625897529535, + "kept_token_count": 573.021875, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896736741065979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0337387690669854e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6947906494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4267822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24000, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30063188076019287, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.13413212522864343, + "grad_norm": 0.00026180720306001604, + "kept_token_count": 572.0734375, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902024328708648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.827189502771944e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6244354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.28961486816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24010, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.3005055636167526, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.13240340054035188, + "grad_norm": 0.00018643711518961936, + "kept_token_count": 572.453125, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903498828411103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7695889113820157e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7978271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28089294433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24020, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.30023619830608367, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.12817247733473777, + "grad_norm": 0.00029225091566331685, + "kept_token_count": 572.6921875, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990024882555008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.896546677424339e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.35224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3364685058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24030, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3001977503299713, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.11358610317111015, + "grad_norm": 0.00018761417595669627, + "kept_token_count": 572.55, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904335856437683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.736895923793782e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2950073242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3074157714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24040, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.29958862364292144, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.12953950464725494, + "grad_norm": 0.00029039589571766555, + "kept_token_count": 573.1890625, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897352039813996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0097138116834687e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7323059082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.32383117675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24050, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2995129257440567, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.12329496964812278, + "grad_norm": 0.0002290268603246659, + "kept_token_count": 573.0125, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904270648956299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.739445601240732e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.30015563964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9872985839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24060, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2999687552452087, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.12406681180000305, + "grad_norm": 0.00023034735932014883, + "kept_token_count": 572.8015625, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901404321193695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.851414239761652e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5978210449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0316650390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24070, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3004645586013794, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.12634214907884597, + "grad_norm": 0.00024904587189666927, + "kept_token_count": 572.5484375, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908232152462005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.584700680221431e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4073028564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0352020263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2995520293712616, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.1249536044895649, + "grad_norm": 0.00018346551223658025, + "kept_token_count": 572.9515625, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904268920421601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.739513304026332e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.53111877441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4776214599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24090, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.300046044588089, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.1262976884841919, + "grad_norm": 0.00019487307872623205, + "kept_token_count": 572.509375, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908198356628418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.586019265640061e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.37889709472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.19240112304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24100, + "valid_token_count": 817.921875 + }, + { + "actual_token_mask_ratio": 0.30014162361621854, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.12503281757235526, + "grad_norm": 0.00015305945998989046, + "kept_token_count": 572.7625, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903089165687561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.785598564718384e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.63583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.33832092285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24110, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3014459252357483, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.12020793631672859, + "grad_norm": 0.00029184407321736217, + "kept_token_count": 571.3703125, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899355411529541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.931443097826559e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.31996154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9346099853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24120, + "valid_token_count": 817.93125 + }, + { + "actual_token_mask_ratio": 0.29912443161010743, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.1382828325033188, + "grad_norm": 0.0002192355168517679, + "kept_token_count": 573.434375, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905468702316285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.692649788717972e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.10230712890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1012908935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24130, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3002756357192993, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.12901028767228126, + "grad_norm": 0.00027059679268859327, + "kept_token_count": 572.571875, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990104615688324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.865407034027157e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0409332275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75881042480466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24140, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.3007191210985184, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.12322475016117096, + "grad_norm": 0.00022388943762052804, + "kept_token_count": 572.053125, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903015673160553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.788464455283247e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1113677978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5621032714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24150, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.29870915710926055, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.1288517877459526, + "grad_norm": 0.00019467499805614352, + "kept_token_count": 573.6953125, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904437720775604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.732924014911987e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.87180786132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.88231201171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24160, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.2995107173919678, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.12826081365346909, + "grad_norm": 0.00025045807706192136, + "kept_token_count": 573.315625, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906769573688508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.641834337031469e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.43016357421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9346008300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.299288210272789, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.1296958528459072, + "grad_norm": 0.00018725069821812212, + "kept_token_count": 573.290625, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902825295925141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7959041947033256e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0449645996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2673645019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3006078481674194, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.1230889305472374, + "grad_norm": 0.00021077244309708476, + "kept_token_count": 572.3359375, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901074230670929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.864307182084303e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9338684082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8189361572266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24190, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.2999340295791626, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.12671620547771453, + "grad_norm": 0.0001896281319204718, + "kept_token_count": 572.8625, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9893670320510864, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.153514637437183e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9690216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.07709350585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24200, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.30116822719573977, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.12576601430773734, + "grad_norm": 0.00024396699154749513, + "kept_token_count": 572.0015625, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900001525878906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.906212295987643e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.54107666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63303527832034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24210, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.30002433657646177, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.12954352423548698, + "grad_norm": 0.00023953379422891885, + "kept_token_count": 572.7609375, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900686800479889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8794368811068126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.93935241699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.18967590332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24220, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.3001704066991806, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.12239483520388603, + "grad_norm": 0.00020645935728680342, + "kept_token_count": 572.46875, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905518651008606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.690697012643795e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8801513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8753204345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24230, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.3010419189929962, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.12448505982756615, + "grad_norm": 0.00018476291734259576, + "kept_token_count": 572.009375, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906599521636963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.648476049420424e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5294677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.91103210449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24240, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3006290256977081, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.12926086634397507, + "grad_norm": 0.00020085854339413345, + "kept_token_count": 572.321875, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900409638881683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.890268271788955e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0351593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6257019042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24250, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.30009604096412656, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.12908345088362694, + "grad_norm": 0.00017810316057875752, + "kept_token_count": 572.684375, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906957745552063, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.634477998275543e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7971710205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3929840087891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24260, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3000153213739395, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.12779682129621506, + "grad_norm": 0.00017982938152272254, + "kept_token_count": 572.9625, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904419779777527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7336242530727756e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8325988769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.60462646484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24270, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.30111436247825624, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.12626724690198898, + "grad_norm": 0.00017919024685397744, + "kept_token_count": 572.059375, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909384489059448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5396872408455235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0246246337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.35045166015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24280, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.30031887590885165, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.13750696778297425, + "grad_norm": 0.00019395006529521197, + "kept_token_count": 572.5609375, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897411644458771, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.007377956440905e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6856964111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5785797119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24290, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.300527286529541, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.12317811995744705, + "grad_norm": 0.00015070743393152952, + "kept_token_count": 572.2453125, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908704400062561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.566258237697184e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0500061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.088720703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24300, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.3010993361473083, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.13079012781381608, + "grad_norm": 0.0002557622210588306, + "kept_token_count": 571.8859375, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908107161521912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.589578900573542e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0124206542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24310, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30011860132217405, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.12869905233383178, + "grad_norm": 0.0004045572131872177, + "kept_token_count": 572.540625, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901003062725067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8670879985147624e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.21863708496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3910766601563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24320, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.2994601517915726, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.1207275502383709, + "grad_norm": 0.0002032546472037211, + "kept_token_count": 572.959375, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906945586204529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6349535548652057e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0569793701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0403717041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24330, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.3007303088903427, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.12656448036432266, + "grad_norm": 0.00019459000031929463, + "kept_token_count": 572.1734375, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910401284694672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4999687704839745e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1179595947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8885894775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24340, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3018586039543152, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.12370126321911812, + "grad_norm": 0.00032292812829837203, + "kept_token_count": 571.2796875, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901636242866516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.842356636596378e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.91957092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.10789489746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24350, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30058847963809965, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.13469803854823112, + "grad_norm": 0.0002328050759388134, + "kept_token_count": 572.2890625, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910146832466126, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5099055276077705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.91962280273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.58592834472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3010331869125366, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.12694922536611558, + "grad_norm": 0.00018952558457385749, + "kept_token_count": 571.8390625, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990497225522995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.71203863323899e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.31621704101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3474914550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24370, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.30002481043338775, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.13194722384214402, + "grad_norm": 0.00028430172824300826, + "kept_token_count": 572.6296875, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904664576053619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7240593883325346e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8792022705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.95457153320314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24380, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.29920895397663116, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.12748443856835365, + "grad_norm": 0.0002763554221019149, + "kept_token_count": 573.5484375, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903907716274262, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7536168747465126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0071624755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0749786376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24390, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29990037679672243, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.12472136989235878, + "grad_norm": 0.000241919988184236, + "kept_token_count": 572.8109375, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901570498943328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.844916573143564e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9632202148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.62086181640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24400, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2994965255260468, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.13753912299871446, + "grad_norm": 0.0002193437103414908, + "kept_token_count": 573.11875, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904051840305328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.747997106984258e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3966369628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.13837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24410, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.3000816464424133, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.12198516353964806, + "grad_norm": 0.0002879235544241965, + "kept_token_count": 572.5171875, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903250455856323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7792947114212436e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3561981201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.77813720703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24420, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.30106047093868255, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.13114674761891365, + "grad_norm": 0.0002556927502155304, + "kept_token_count": 572.0890625, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901719331741333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8391061752918174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2746887207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43951721191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24430, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.29989940524101255, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.12721626907587053, + "grad_norm": 0.00022942798386793584, + "kept_token_count": 572.7, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903749108314515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.759818100661505e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6305725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5152526855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24440, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2998463809490204, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.1384018398821354, + "grad_norm": 0.00019883876666426659, + "kept_token_count": 573.00625, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905032396316529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7096862433827484e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5538360595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24351806640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24450, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.30027277767658234, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.12397703975439071, + "grad_norm": 0.00020817146287299693, + "kept_token_count": 572.29375, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907593369483948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6096558687859215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5977813720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7137512207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24460, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.29829190373420716, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.1261409915983677, + "grad_norm": 0.0001897752663353458, + "kept_token_count": 574.1046875, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908122301101685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5889910941477864e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0373016357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.52826232910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24470, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29999050199985505, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.13628939911723137, + "grad_norm": 0.0001749281509546563, + "kept_token_count": 572.8578125, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906036615371704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.670463902381016e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1061737060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6371215820312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24480, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.30167570114135744, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.127132660895586, + "grad_norm": 0.00015487978816963732, + "kept_token_count": 571.3484375, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907462418079376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.614778397604823e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.40284118652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1524353027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.30065118372440336, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.12917086780071257, + "grad_norm": 0.00023785774828866124, + "kept_token_count": 572.2203125, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903014123439788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7885261190240274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4954345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31505126953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24500, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2992906779050827, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.1316301070153713, + "grad_norm": 0.0001938357308972627, + "kept_token_count": 573.44375, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990520465373993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.702967696881387e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.13674011230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7545684814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24510, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30095010101795194, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.1235727347433567, + "grad_norm": 0.00021316336642485112, + "kept_token_count": 572.190625, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905152022838593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.705012422869913e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1413818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.25946350097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24520, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.3002363383769989, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.11966729164123535, + "grad_norm": 0.0003008627681992948, + "kept_token_count": 572.515625, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909570276737213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.532433420332382e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.812890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.84679260253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.3005527347326279, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.1389629878103733, + "grad_norm": 0.00021443743025884032, + "kept_token_count": 572.5078125, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907619535923005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608628812798997e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4760375976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0737609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24540, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.29980230033397676, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.1243108369410038, + "grad_norm": 0.0001602226693648845, + "kept_token_count": 572.965625, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902491569519043, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.808941728493665e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.80616149902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0917694091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24550, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.30029005408287046, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.13249293193221093, + "grad_norm": 0.00027732228045351803, + "kept_token_count": 572.428125, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902602732181549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.804608240898233e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5212768554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3247131347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24560, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3002225250005722, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.13512909635901452, + "grad_norm": 0.00025039887987077236, + "kept_token_count": 572.721875, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908031225204468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5925485644838775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.68193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6258880615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24570, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.30022641122341154, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.12199540361762047, + "grad_norm": 0.0002698977186810225, + "kept_token_count": 572.6609375, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903082728385926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7858490759390404e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.07061157226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.61989135742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24580, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.300164532661438, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.12200800999999047, + "grad_norm": 0.0002988535270560533, + "kept_token_count": 572.61875, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901965260505676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8294964542728846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.02223510742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9088562011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24590, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29963418245315554, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.12907756567001344, + "grad_norm": 0.00020582832803484052, + "kept_token_count": 573.1515625, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906648516654968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.646563345682807e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.53196411132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2379089355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24600, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.3003024458885193, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.12822262048721314, + "grad_norm": 0.0002074100630125031, + "kept_token_count": 572.575, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910822749137879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.483512737147976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.17014770507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9673156738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24610, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.29971430003643035, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.1278340071439743, + "grad_norm": 0.00024465552996844053, + "kept_token_count": 572.8703125, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905887007713318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.676305732369656e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7094390869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9033599853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24620, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.30051265060901644, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.12566427513957024, + "grad_norm": 0.00018052459927275777, + "kept_token_count": 572.328125, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908630788326264, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.569132604752667e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0248229980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6885070800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24630, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999788224697113, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.11624064072966575, + "grad_norm": 0.00022175651974976063, + "kept_token_count": 572.884375, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908256769180298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5837387076753655e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.470458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.78785095214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24640, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.30064627528190613, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.11651663556694984, + "grad_norm": 0.0002024866989813745, + "kept_token_count": 572.0125, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909009575843811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5543405101634565e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.68623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1775634765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24650, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.29892543852329256, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.13877382799983023, + "grad_norm": 0.0002134526876034215, + "kept_token_count": 573.6640625, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990597379207611, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.672919228847604e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.14390258789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2433349609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.29952734112739565, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.1231289729475975, + "grad_norm": 0.00018837369862012565, + "kept_token_count": 573.1125, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990671569108963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6439394534681924e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6596160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6445587158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24670, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29967125654220583, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.12266210317611695, + "grad_norm": 0.000287075323285535, + "kept_token_count": 572.9453125, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903204381465912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7810975845786744e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.61668090820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08782653808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24680, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2997272700071335, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.13371036052703858, + "grad_norm": 0.00021174203720875084, + "kept_token_count": 573.225, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897871673107147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.989408760389779e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1213012695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3395080566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24690, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.29993718564510347, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.12208429127931594, + "grad_norm": 0.0002224613563157618, + "kept_token_count": 572.746875, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906566381454468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6497762448561846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.85350646972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1185272216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24700, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.30090347230434417, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.13415421918034554, + "grad_norm": 0.0002475746732670814, + "kept_token_count": 571.88125, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897522330284119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0030540912994186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3449951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5088745117188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24710, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.30054869055747985, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.12711765021085739, + "grad_norm": 0.0001942821400007233, + "kept_token_count": 572.2171875, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898717939853668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.95635546738049e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1080383300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6375030517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2994673877954483, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.11559270843863487, + "grad_norm": 0.00027098788996227086, + "kept_token_count": 573.2875, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904074788093566, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.747095106518827e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3289306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6343658447266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24730, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.29966809749603274, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.13187507092952727, + "grad_norm": 0.00020017407950945199, + "kept_token_count": 573.3125, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905730962753296, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.682406568259466e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0711212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3670959472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24740, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.29930195212364197, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.1218710720539093, + "grad_norm": 0.00022499097394756973, + "kept_token_count": 573.66875, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898047685623169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.98253101593582e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.65390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1928985595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24750, + "valid_token_count": 818.7125 + }, + { + "actual_token_mask_ratio": 0.30007403194904325, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.12883778661489487, + "grad_norm": 0.00020510506874416023, + "kept_token_count": 572.559375, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903366267681122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.774773213081062e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3804931640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.62164611816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24760, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29917733669281005, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.13522468656301498, + "grad_norm": 0.0003053275286220014, + "kept_token_count": 573.5890625, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905483245849609, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6920838101650586e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8401733398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8058746337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24770, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29985483586788175, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.12230328097939491, + "grad_norm": 0.0002110224450007081, + "kept_token_count": 573.078125, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907577633857727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.610265375755262e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3292755126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.88306884765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24780, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2993565917015076, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.13007266223430633, + "grad_norm": 0.00021553630358539522, + "kept_token_count": 573.409375, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899859368801117, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.911761596100405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1325622558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1872985839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24790, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.29938446283340453, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.1307036705315113, + "grad_norm": 0.00020625715842470527, + "kept_token_count": 573.3640625, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905205309391022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.702932845044416e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0003265380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.18818359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24800, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3001782864332199, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.12900728359818459, + "grad_norm": 0.00020752710406668484, + "kept_token_count": 572.775, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906111359596252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6675437149824576e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.46251220703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3258483886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24810, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.301216185092926, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.12774882465600967, + "grad_norm": 0.00030548160430043936, + "kept_token_count": 571.9421875, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903571367263794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7667634387617e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.53387451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.07435913085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24820, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.29916841089725493, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.12805298939347268, + "grad_norm": 0.0002035013458225876, + "kept_token_count": 573.371875, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904726564884185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.721634711837396e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2143615722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6022918701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24830, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.29954342246055604, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.1232452318072319, + "grad_norm": 0.0002818548528011888, + "kept_token_count": 573.0671875, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905419051647186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6945864849258214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8428070068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9513732910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24840, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29883414804935454, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.1321046233177185, + "grad_norm": 0.00023608366609551013, + "kept_token_count": 573.6515625, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905705749988556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6833883496001366e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.38002014160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75583801269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24850, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.29992911517620086, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.1369418039917946, + "grad_norm": 0.00023543077986687422, + "kept_token_count": 573.05625, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896600902080536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.039044142700732e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0489440917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.60369873046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24860, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.3005161195993423, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.12675570398569108, + "grad_norm": 0.00021370724425651133, + "kept_token_count": 572.490625, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900985598564148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.867771520162933e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4365753173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9572326660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24870, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.29914380609989166, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.12881070449948312, + "grad_norm": 0.00019301622523926198, + "kept_token_count": 573.4421875, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905445277690887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.693567850859836e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7340087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3417541503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24880, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3015361547470093, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.12711478397250175, + "grad_norm": 0.00021315517369657755, + "kept_token_count": 571.671875, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902623772621155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8037795457057654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8855834960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9330780029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24890, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.30026707649230955, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.11370129212737083, + "grad_norm": 0.0002290195261593908, + "kept_token_count": 572.740625, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900481402873993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8874634992680515e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.128564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.96611328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24900, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2998971164226532, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.13252585232257844, + "grad_norm": 0.00022430448734667152, + "kept_token_count": 572.971875, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910535752773285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.494717857392971e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.411083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1929138183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24910, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2996643453836441, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.13571734577417374, + "grad_norm": 0.00018648534023668617, + "kept_token_count": 573.2015625, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902795016765594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.797089266299736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4516265869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.68699645996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24920, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.3000515758991241, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.1266350768506527, + "grad_norm": 0.00019273163343314081, + "kept_token_count": 572.884375, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905653536319733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6854259451502e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7742431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06760864257814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24930, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2992131680250168, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.1335723429918289, + "grad_norm": 0.00015897324192337692, + "kept_token_count": 573.296875, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904445052146912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7326361052691936e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.80907287597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.94622802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24940, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.30189151763916017, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.12220175191760063, + "grad_norm": 0.0002198973816120997, + "kept_token_count": 571.1765625, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908090829849243, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5902207309845834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5203369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9180450439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24950, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.30095568001270295, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.13115599676966666, + "grad_norm": 0.00020225663320161402, + "kept_token_count": 571.96875, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905463218688965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.692862119351048e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1470581054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.32147521972655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24960, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30060626566410065, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.12732433825731276, + "grad_norm": 0.0002515266533009708, + "kept_token_count": 572.2515625, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899911880493164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.909712759195827e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3577453613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3166931152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3009146511554718, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.12544939965009688, + "grad_norm": 0.00021562450274359435, + "kept_token_count": 571.8703125, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910070538520813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5128866147715596e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7933624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6507843017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24980, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2997991979122162, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.11284269839525223, + "grad_norm": 0.00022815300326328725, + "kept_token_count": 572.7890625, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910339117050171, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.502399358694675e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8042266845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.3177185058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24990, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.29968503415584563, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.13791835755109788, + "grad_norm": 0.00020038828370161355, + "kept_token_count": 573.128125, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990831333398819, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5815337105304935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5921234130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.21229248046876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25000, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2996665984392166, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.12393337860703468, + "grad_norm": 0.00029550225008279085, + "kept_token_count": 573.0015625, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904912531375885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.714367703651078e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8356903076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7278594970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25010, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2977187991142273, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.1259586475789547, + "grad_norm": 0.00027044108719564974, + "kept_token_count": 574.628125, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903636276721954, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.764222674362827e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.92002258300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4577117919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25020, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3005314886569977, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.13885861486196518, + "grad_norm": 0.00016920773487072438, + "kept_token_count": 572.0671875, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896692097187042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.035481579194311e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8652587890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.15013732910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25030, + "valid_token_count": 817.859375 + }, + { + "actual_token_mask_ratio": 0.30015942454338074, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.12741965427994728, + "grad_norm": 0.00025418741279281676, + "kept_token_count": 572.6640625, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908282399177551, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5827456304104996e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1550720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7535034179688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25040, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3000086843967438, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.11869213208556176, + "grad_norm": 0.00020955938089173287, + "kept_token_count": 572.6453125, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908368408679962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5793785355053845e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0565704345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.11451721191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25050, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2992178052663803, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.13191300854086876, + "grad_norm": 0.00023669730580877513, + "kept_token_count": 573.2203125, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906053662300109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6697984614875165e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.56810302734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9557586669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25060, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.2994889795780182, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.1260331690311432, + "grad_norm": 0.0003529581008478999, + "kept_token_count": 573.121875, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898665487766266, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.958394736400805e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7973663330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9861785888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25070, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.3007655918598175, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.13423280119895936, + "grad_norm": 0.0001996503706322983, + "kept_token_count": 572.0671875, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0, + "mgd_cos_sim": 0.991012179851532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.510882052069064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4925048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.74835205078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29896190464496614, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.13445331305265426, + "grad_norm": 0.00023351750860456377, + "kept_token_count": 573.6921875, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907171726226807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.626123689173255e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.33449401855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.71692810058596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25090, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30036230087280275, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.12046372145414352, + "grad_norm": 0.0004018802137579769, + "kept_token_count": 572.5328125, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904395580291748, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.734564852493349e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.42137145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1755676269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25100, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29996952712535857, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.11925019696354866, + "grad_norm": 0.00022941695351619273, + "kept_token_count": 572.78125, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909613490104675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5307415964780374e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8722259521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3182373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25110, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30059259235858915, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.12321283593773842, + "grad_norm": 0.0003265286504756659, + "kept_token_count": 572.21875, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990509444475174, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.707268224388827e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.004248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4185485839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25120, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.2999707251787186, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.13207819163799286, + "grad_norm": 0.0003431648947298527, + "kept_token_count": 572.740625, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899827420711518, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.913011569238734e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.33363037109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3269897460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25130, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2992462158203125, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.14258537963032722, + "grad_norm": 0.00020046539430040866, + "kept_token_count": 573.16875, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904522895812988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.729590498551261e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9612701416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.18848571777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25140, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.29977098703384397, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.11827405914664268, + "grad_norm": 0.0002507095050532371, + "kept_token_count": 573.0453125, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990641063451767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.655852560768835e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.27094116210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.06676330566404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25150, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2991989880800247, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.13197021633386613, + "grad_norm": 0.00017950795881915838, + "kept_token_count": 573.740625, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908012688159943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5932772152591494e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7489074707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0992401123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25160, + "valid_token_count": 818.696875 + }, + { + "actual_token_mask_ratio": 0.29994619488716123, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.12261785566806793, + "grad_norm": 0.00025344640016555786, + "kept_token_count": 572.63125, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900419175624847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.889894433086738e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1232635498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8060791015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.3000400960445404, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.13434277027845382, + "grad_norm": 0.00033276950125582516, + "kept_token_count": 572.9203125, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905978322029114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6727422957483215e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0396789550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.098046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25180, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.30046975910663604, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.1296377435326576, + "grad_norm": 0.00025629400624893606, + "kept_token_count": 572.2921875, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909192144870758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5471969022182746e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3238983154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5614990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25190, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29963410198688506, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.12830770313739776, + "grad_norm": 0.00020238237630110234, + "kept_token_count": 573.20625, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902229845523834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.819167432084214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.44500427246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.227392578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25200, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.3003549337387085, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.13196595162153243, + "grad_norm": 0.0002713683934416622, + "kept_token_count": 572.4546875, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901603519916534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.84362785553094e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6942901611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17270812988284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25210, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3012383669614792, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.1257832407951355, + "grad_norm": 0.00021195277804508805, + "kept_token_count": 571.6, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909425735473633, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.538082655722974e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.879736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.2954620361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25220, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3000020861625671, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.12020564079284668, + "grad_norm": 0.00025400458252988756, + "kept_token_count": 572.5796875, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903903603553772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.753788405447267e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.10740661621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.6494384765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25230, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.2993404924869537, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.12107751592993736, + "grad_norm": 0.00018406535673420876, + "kept_token_count": 573.2125, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911331593990326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.463624834694201e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0334167480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.615087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25240, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.301088348031044, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.13209532499313353, + "grad_norm": 0.00023592673824168742, + "kept_token_count": 571.8890625, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905356585979461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.69703040632885e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.55548706054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.51256103515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25250, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30055407285690305, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.13066064268350602, + "grad_norm": 0.00022663807612843812, + "kept_token_count": 572.11875, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900134563446045, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.90101587981917e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5629089355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.37445373535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25260, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.30018840730190277, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.12641912177205086, + "grad_norm": 0.00019861088367179036, + "kept_token_count": 572.759375, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990554255247116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.689766963361762e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.88352661132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9756103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25270, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.299906975030899, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.12086287662386894, + "grad_norm": 0.00032603717409074306, + "kept_token_count": 573.0078125, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900490641593933, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.887106940965168e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.79566650390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.73253173828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25280, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.2991274118423462, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.12925286665558816, + "grad_norm": 0.00020341693016234785, + "kept_token_count": 573.171875, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906456112861634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.654079519037623e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.45668640136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4085693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25290, + "valid_token_count": 817.7984375 + }, + { + "actual_token_mask_ratio": 0.2999181360006332, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.12435556873679161, + "grad_norm": 0.0002017249062191695, + "kept_token_count": 572.7890625, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902758121490478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.798528068728046e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9528442382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.85462646484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.300407800078392, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.12888087704777718, + "grad_norm": 0.00026180557324551046, + "kept_token_count": 572.70625, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904867053031922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7161557702347634e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.02715454101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.35924072265624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25310, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.299177947640419, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.12670276388525964, + "grad_norm": 0.00019223291019443423, + "kept_token_count": 573.6109375, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911637902259827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.451666088949423e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2371826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4806854248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25320, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.3010212302207947, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.13227966651320458, + "grad_norm": 0.00021595333237200975, + "kept_token_count": 572.1015625, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990279632806778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7970393350406084e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3413055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.99342651367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25330, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.30022579431533813, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.13025395199656487, + "grad_norm": 0.0001882296201074496, + "kept_token_count": 572.3015625, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990416431427002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.743608067452442e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.59254150390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4763977050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25340, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.29909609258174896, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.1311442345380783, + "grad_norm": 0.0002319774212082848, + "kept_token_count": 573.475, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990967434644699, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.528367287799483e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.38905639648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3325988769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.30081455409526825, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.12548589929938317, + "grad_norm": 0.00021757899958174676, + "kept_token_count": 572.1953125, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898831725120545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.951901417167392e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.105712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9061676025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25360, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2996148467063904, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.13448756337165832, + "grad_norm": 0.00023453669564332813, + "kept_token_count": 573.0875, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9909725427627564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5263672180008145e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5386688232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8015991210938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25370, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30124208629131316, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.12910555824637412, + "grad_norm": 0.0002773674495983869, + "kept_token_count": 571.7859375, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901885032653809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.832635011349339e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.63483276367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3313751220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25380, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2983100175857544, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.12414919808506966, + "grad_norm": 0.00019951809372287244, + "kept_token_count": 574.0796875, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899667739868164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.919249720638618e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0869842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1043975830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25390, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2988921731710434, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.12372457087039948, + "grad_norm": 0.00022984603128861636, + "kept_token_count": 573.7359375, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904838681221009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.717255494848359e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6296905517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.92755737304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25400, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29999835789203644, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.1359150841832161, + "grad_norm": 0.0002058387326542288, + "kept_token_count": 572.78125, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900076448917389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.903280739905313e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.60663146972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5189453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25410, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29916237890720365, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.13851595297455788, + "grad_norm": 0.0002010370371863246, + "kept_token_count": 573.7421875, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908458173274994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.575869050109759e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.42919006347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5419555664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25420, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.30067568421363833, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.1256292723119259, + "grad_norm": 0.00030498573323711753, + "kept_token_count": 572.2859375, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903943240642548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.752233897102997e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.34861450195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4285003662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25430, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2996579945087433, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.1246878907084465, + "grad_norm": 0.0002076951932394877, + "kept_token_count": 572.98125, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902042031288147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8264971954049544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0640441894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.43458557128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25440, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.29996836483478545, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.12516569048166276, + "grad_norm": 0.0003574857546482235, + "kept_token_count": 573.1328125, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904242217540741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.740557258424815e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6063293457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7434417724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25450, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.3002841234207153, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.12677211612462996, + "grad_norm": 0.00024716739426366985, + "kept_token_count": 572.5609375, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907696127891541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6056355747859924e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.29982604980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8816864013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25460, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30020139515399935, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.13297978788614273, + "grad_norm": 0.00018589805404189974, + "kept_token_count": 572.590625, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903814494609833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.75726303900592e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4971099853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5673126220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25470, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2996331244707108, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.12904790565371513, + "grad_norm": 0.0002051869814749807, + "kept_token_count": 572.996875, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904000103473664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.750013856915757e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3433868408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1167938232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25480, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2991791933774948, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.12851145640015602, + "grad_norm": 0.0002489452890586108, + "kept_token_count": 573.2578125, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904001772403717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7499535574170294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.90651245117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2960968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25490, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2998009204864502, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.1326180227100849, + "grad_norm": 0.0002146430779248476, + "kept_token_count": 573.0890625, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905090391635895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.707423202286009e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.16261901855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6156890869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3003909349441528, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.12407935187220573, + "grad_norm": 0.00024111093080136925, + "kept_token_count": 572.534375, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907996177673339, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5939135341322984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7538360595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0680877685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25510, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30039929449558256, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.12330802977085113, + "grad_norm": 0.0001937639171956107, + "kept_token_count": 572.675, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901670753955841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.841001889668405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3452880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.22535095214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25520, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3006279170513153, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.12412323504686355, + "grad_norm": 0.0002105387393385172, + "kept_token_count": 572.2421875, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905673444271088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.684646944748238e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7685485839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5027557373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25530, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2992181181907654, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.1366240970790386, + "grad_norm": 0.0002222585171693936, + "kept_token_count": 573.4828125, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905619859695435, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.686744312290102e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2744995117188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25540, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.29976622462272645, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.13015155345201493, + "grad_norm": 0.00015600032929796726, + "kept_token_count": 572.9671875, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902798891067505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7969400000292805e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8161865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.66148681640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25550, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3009325087070465, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.12784079611301422, + "grad_norm": 0.00030677238828502595, + "kept_token_count": 572.1765625, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904294669628143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7385082578111906e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4490936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.02276306152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25560, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.30034137666225436, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.12770074978470802, + "grad_norm": 0.00030234039877541363, + "kept_token_count": 572.4671875, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990509831905365, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7071203405503186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.54967041015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.23578491210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25570, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30017662346363067, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.12459172755479812, + "grad_norm": 0.0002756736648734659, + "kept_token_count": 572.79375, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903327047824859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7763021828141065e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3708129882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.48320922851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25580, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.3002544194459915, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.13435222804546357, + "grad_norm": 0.00021303980611264706, + "kept_token_count": 572.596875, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905358791351319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.696935782500077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.3998321533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.63930053710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25590, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2995900422334671, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.12879453748464584, + "grad_norm": 0.0003019057330675423, + "kept_token_count": 573.3578125, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903288364410401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.777816982619697e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.15825805664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.27227783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25600, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.30080884993076323, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.13113659173250197, + "grad_norm": 0.00022767182963434607, + "kept_token_count": 571.9734375, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989968991279602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9183845365187155e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3036346435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.27553100585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25610, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.3003801256418228, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.12162097394466401, + "grad_norm": 0.00019110106222797185, + "kept_token_count": 572.571875, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906047940254211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.670019577839412e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3656036376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2929626464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30097064971923826, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.13060771375894548, + "grad_norm": 0.0002025335852522403, + "kept_token_count": 572.1921875, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906170189380645, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.665246731543448e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3539123535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08848266601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25630, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.29918323159217836, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.12191623374819756, + "grad_norm": 0.0002381417725700885, + "kept_token_count": 573.259375, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904222309589386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.741330219781957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.73044738769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.84978637695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25640, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.29927551448345185, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.13212104067206382, + "grad_norm": 0.00018057311535812914, + "kept_token_count": 573.36875, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897754728794098, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9939796988619494e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7165466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.10716857910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25650, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.3002162009477615, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.12232114598155022, + "grad_norm": 0.00020143878646194935, + "kept_token_count": 572.6859375, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900646448135376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.881016018567607e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6287048339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1496154785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25660, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3010080337524414, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.13179056346416473, + "grad_norm": 0.0002416540082776919, + "kept_token_count": 572.00625, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910149574279785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.509798989398405e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9254974365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.51604919433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25670, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30001468062400816, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.13066897615790368, + "grad_norm": 0.0001797986769815907, + "kept_token_count": 572.58125, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902770042419433, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7980628621880894e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.42755432128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65162353515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25680, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.3004977762699127, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.12457880899310111, + "grad_norm": 0.0002083161671180278, + "kept_token_count": 572.5875, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901780307292938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.836722498817835e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.93079833984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.8985534667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25690, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.29966007471084594, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.12978836819529532, + "grad_norm": 0.0003475802659522742, + "kept_token_count": 573.003125, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904548823833466, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7285742109816056e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3437927246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0510650634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25700, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.30001455545425415, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.12954277247190477, + "grad_norm": 0.00021052156807854772, + "kept_token_count": 572.5109375, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902947425842286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7911290928605015e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0210815429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.02220153808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25710, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.30056331157684324, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.1343002051115036, + "grad_norm": 0.00020124964066781104, + "kept_token_count": 572.496875, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905644178390502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.685786396090407e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7772491455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5831024169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25720, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.3003310769796371, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.11734918504953384, + "grad_norm": 0.00018598248425405473, + "kept_token_count": 572.4234375, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907609045505523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.609042760217562e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5240936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8770751953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25730, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2998334109783173, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.12414917796850204, + "grad_norm": 0.00020526062871795148, + "kept_token_count": 572.7640625, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906001091003418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.671847844088916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.60321350097655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.66249084472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25740, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30032520592212675, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.12187121212482452, + "grad_norm": 0.0001862632780103013, + "kept_token_count": 572.3921875, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9901861429214478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.833558912447188e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7671630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.46119689941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25750, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.30048151314258575, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.11723266020417214, + "grad_norm": 0.00019928206165786833, + "kept_token_count": 572.425, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903436899185181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7720146428910085e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.19854125976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1506042480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25760, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2998168408870697, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.1258074790239334, + "grad_norm": 0.0001959805958904326, + "kept_token_count": 572.921875, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902243137359619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.818654986389447e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3170593261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7412567138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2999183744192123, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.12806259989738464, + "grad_norm": 0.00022805953631177545, + "kept_token_count": 572.790625, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906749367713928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6426269434741695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.212548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.542236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25780, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29909308850765226, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.13155997022986413, + "grad_norm": 0.00019985632388852537, + "kept_token_count": 573.421875, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903910040855408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.753531746042427e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.39471435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.61902770996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25790, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3007842928171158, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.12261410132050514, + "grad_norm": 0.00018391454068478197, + "kept_token_count": 571.971875, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902323603630065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8155046058818695e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4113433837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0301574707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25800, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3003565281629562, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.12999649196863175, + "grad_norm": 0.00020203222811687738, + "kept_token_count": 572.5375, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990561717748642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6868444658466616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.66151123046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9402648925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25810, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30119763910770414, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.1334328219294548, + "grad_norm": 0.0002420207456452772, + "kept_token_count": 571.7484375, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904603064060211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.726463291968685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.21751708984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.13169555664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25820, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.3002035588026047, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.12428395599126815, + "grad_norm": 0.00019348599016666412, + "kept_token_count": 572.746875, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989851975440979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9640925569983666e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.04522705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4597442626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25830, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.30117634534835813, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.13349199891090394, + "grad_norm": 0.00030371363391168416, + "kept_token_count": 571.8734375, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904173314571381, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.743254383152817e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5715026855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.98484802246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25840, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.30059911608695983, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.11504823341965675, + "grad_norm": 0.0001796534488676116, + "kept_token_count": 572.5, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989862585067749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.959937603212893e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9204498291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0793518066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25850, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.30036342740058897, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.1282852567732334, + "grad_norm": 0.00019746382895391434, + "kept_token_count": 572.4765625, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906508207321167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.652042178146075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4101287841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.34641418457034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25860, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30068681240081785, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.13472290113568305, + "grad_norm": 0.0002587883791420609, + "kept_token_count": 572.503125, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899019777774811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.944564159610309e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.12444763183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.58154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25870, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.2997268795967102, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.12232587561011314, + "grad_norm": 0.00042834371561184525, + "kept_token_count": 573.015625, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902466297149658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.809927329712082e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.24701538085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1053527832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29971787333488464, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.13695763275027276, + "grad_norm": 0.0001941337250173092, + "kept_token_count": 572.7921875, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906154453754425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6658638055087064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8463928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2776824951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25890, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.30003180503845217, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.1311656765639782, + "grad_norm": 0.0002646333887241781, + "kept_token_count": 572.6703125, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903746247291565, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.759930732485373e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.87352600097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20106201171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25900, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2996992409229279, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.12631291672587394, + "grad_norm": 0.00021893253142479807, + "kept_token_count": 572.9125, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906436264514923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.65485371730756e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.76153564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3909210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25910, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.29987495243549345, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.12064834237098694, + "grad_norm": 0.00022020432516001165, + "kept_token_count": 573.08125, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904589295387268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.727000039361883e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5716583251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.16875915527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25920, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.3002647072076797, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.12654675394296647, + "grad_norm": 0.0002004072885029018, + "kept_token_count": 572.35, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908293545246124, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5822977952193466e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3390747070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.34897766113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25930, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.30098487734794616, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.12411445006728172, + "grad_norm": 0.00024120612943079323, + "kept_token_count": 571.8265625, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904747605323792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7208155117696154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0226806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.55039367675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25940, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.29964114129543307, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.12629991695284842, + "grad_norm": 0.00019074644660577178, + "kept_token_count": 573.06875, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904876530170441, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7157774931984024e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.17779846191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5261535644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25950, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30039710402488706, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.1268260046839714, + "grad_norm": 0.0002121076686307788, + "kept_token_count": 572.684375, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904063522815705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.747540304175345e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.08397521972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7894714355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25960, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.3009284406900406, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.12970836088061333, + "grad_norm": 0.00022120728681329638, + "kept_token_count": 572.1984375, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908072292804718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5909455255023205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8539184570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4662170410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25970, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.3007729709148407, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.13704759776592254, + "grad_norm": 0.00015974264533724636, + "kept_token_count": 572.2875, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903255999088287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7790807982673866e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6073059082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4219604492188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25980, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.299025821685791, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.13168938755989074, + "grad_norm": 0.0002048315218416974, + "kept_token_count": 573.7484375, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989782041311264, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.991408048023004e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.94146423339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25990, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2991581380367279, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.12726994305849076, + "grad_norm": 0.0001653713989071548, + "kept_token_count": 573.3171875, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899808466434479, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9137549174483864e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6880401611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36322631835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26000, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.3005285322666168, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.1299401342868805, + "grad_norm": 0.000270975346211344, + "kept_token_count": 572.296875, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9894020855426788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.1398304529138843e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6390838623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.41734313964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2997519701719284, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.1285797782242298, + "grad_norm": 0.0002255227736895904, + "kept_token_count": 572.946875, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9897342979907989, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0100571641232816e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.07602844238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.88172302246096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3003548264503479, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.12556107789278032, + "grad_norm": 0.0003210343129467219, + "kept_token_count": 572.8171875, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900886237621307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8716501512681135e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.38294982910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2509063720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26030, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.30055544078350066, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.13398505002260208, + "grad_norm": 0.000213300678296946, + "kept_token_count": 572.05625, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902340471744537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.814847550529521e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3932708740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3069274902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26040, + "valid_token_count": 817.8765625 + }, + { + "actual_token_mask_ratio": 0.29952462315559386, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.12333126738667488, + "grad_norm": 0.00023453624453395605, + "kept_token_count": 573.1390625, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902737557888031, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.799332662310917e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6766326904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5604217529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26050, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.30080935060977937, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.1345680557191372, + "grad_norm": 0.00019402196630835533, + "kept_token_count": 572.0421875, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9908933937549591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5572900560509876e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4953948974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7426483154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26060, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.30030105710029603, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.12547452449798585, + "grad_norm": 0.0001880488416645676, + "kept_token_count": 572.784375, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906386256217956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6568072209774984e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5463409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.38844604492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26070, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.3005603730678558, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.1185310497879982, + "grad_norm": 0.0002040885156020522, + "kept_token_count": 572.046875, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906818389892578, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.639925344032235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.33604431152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8561706542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26080, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.3006723403930664, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.12531975284218788, + "grad_norm": 0.0002013526827795431, + "kept_token_count": 572.29375, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990478229522705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.719466512848157e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.23322143554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.18889770507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26090, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3008009195327759, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.1258675254881382, + "grad_norm": 0.0001842768833739683, + "kept_token_count": 572.2984375, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903838753700256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7563106889137996e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.42969665527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.93373413085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26100, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.2995903342962265, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.13738378509879112, + "grad_norm": 0.00020850454166065902, + "kept_token_count": 573.18125, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904266119003295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.739619060070254e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.48101196289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8597412109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26110, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.30132928490638733, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.13038603439927102, + "grad_norm": 0.00024214203585870564, + "kept_token_count": 571.671875, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903486609458924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.770068469748367e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3362274169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7959747314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26120, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2997540235519409, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.13065618500113488, + "grad_norm": 0.00018306580022908747, + "kept_token_count": 572.884375, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902735769748687, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.799397200054955e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8380157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5083740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26130, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29926698803901675, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.12950392439961433, + "grad_norm": 0.00024513862445019186, + "kept_token_count": 573.3234375, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902036249637604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8267226773314175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1932434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.36305236816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.29992492198944093, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.12796334698796272, + "grad_norm": 0.0001802898186724633, + "kept_token_count": 572.834375, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902388036251069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.812992235907586e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0852844238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9635925292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26150, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.3008978307247162, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.12358767837285996, + "grad_norm": 0.00035902406671084464, + "kept_token_count": 571.7953125, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9896621942520142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.038220940856263e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1309875488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.21275939941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26160, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.298656964302063, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.11889718845486641, + "grad_norm": 0.0002692010602913797, + "kept_token_count": 574.1453125, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907850921154022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5995965845359024e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.22752990722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.70252990722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26170, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.29965344071388245, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.1268678553402424, + "grad_norm": 0.0002193280670326203, + "kept_token_count": 572.8671875, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903172731399537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.782331532420358e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.76031494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1863677978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26180, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.29983951449394225, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.1269281215965748, + "grad_norm": 0.0002424699196126312, + "kept_token_count": 572.5859375, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9898509740829468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9644800199312155e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0874542236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.90916137695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26190, + "valid_token_count": 817.7875 + }, + { + "actual_token_mask_ratio": 0.2981178969144821, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.1259082540869713, + "grad_norm": 0.00032275839475914836, + "kept_token_count": 574.496875, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989979088306427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.914439075742848e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9326477050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8669067382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26200, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.29955037236213683, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.12319510579109191, + "grad_norm": 0.00018970161909237504, + "kept_token_count": 573.240625, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905313849449158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6986975828767754e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3230712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.85003967285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26210, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29921689033508303, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.12461930736899376, + "grad_norm": 0.0002600968873593956, + "kept_token_count": 573.5046875, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0, + "mgd_cos_sim": 0.989963948726654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.920355302398093e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1205627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3522918701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26220, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.29944397509098053, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.13286631777882577, + "grad_norm": 0.00026933784829452634, + "kept_token_count": 573.20625, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899694502353669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9182013279059905e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.201953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.22243041992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3002632588148117, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.12371915876865387, + "grad_norm": 0.00021890514472033828, + "kept_token_count": 572.3953125, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910577893257141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4930655237985776e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2630584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7115203857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26240, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3001388281583786, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.12012575939297676, + "grad_norm": 0.00023203575983643532, + "kept_token_count": 572.69375, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906700372695922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.644540374807548e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.14255676269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.48167724609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26250, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.30003381371498106, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.13184379041194916, + "grad_norm": 0.00020589808991644531, + "kept_token_count": 572.75, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906840741634368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6390524110174737e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3981567382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5585174560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26260, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2997371554374695, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.12253862246870995, + "grad_norm": 0.00022532397997565567, + "kept_token_count": 573.0546875, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904491603374481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.730814787559211e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6054748535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0388488769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3003206968307495, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.12499450445175171, + "grad_norm": 0.0002445752907078713, + "kept_token_count": 572.58125, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899911880493164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.909712286258582e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.34202575683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26280, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.29935245513916015, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.1316678337752819, + "grad_norm": 0.00018077252025250345, + "kept_token_count": 573.0046875, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906091392040253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.668321987788658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.81573486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26290, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.3005318731069565, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.11712622046470642, + "grad_norm": 0.00023250504455063492, + "kept_token_count": 572.70625, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906469225883484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.653565509011969e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3420379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.58241271972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26300, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.300540566444397, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.12732767388224603, + "grad_norm": 0.00018131668912246823, + "kept_token_count": 572.571875, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910123944282532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.51080179825658e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.84222412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.864453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26310, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.2993190586566925, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.12262820601463317, + "grad_norm": 0.00027074693934991956, + "kept_token_count": 573.346875, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900076925754547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.90326778870076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3364715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1371826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26320, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29999746680259703, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.13021026849746703, + "grad_norm": 0.0001901632931549102, + "kept_token_count": 573.215625, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910574376583099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.493214662739774e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.28045043945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.59423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26330, + "valid_token_count": 818.8765625 + }, + { + "actual_token_mask_ratio": 0.3006333470344543, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.13740539476275443, + "grad_norm": 0.0001821624900912866, + "kept_token_count": 572.290625, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903918266296386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7532136411755344e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.98529357910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2095642089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26340, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.29933039247989657, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.12251203507184982, + "grad_norm": 0.00020051556930411607, + "kept_token_count": 573.1359375, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906532287597656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6510994323180054e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3980010986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.52123107910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26350, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.30043174624443053, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.12683407738804817, + "grad_norm": 0.00020052483887411654, + "kept_token_count": 572.3390625, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9903049528598785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7871428503422067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6718780517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.26796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26360, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2999173104763031, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.1349111430346966, + "grad_norm": 0.00030653004068881273, + "kept_token_count": 572.95, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906000316143035, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.671881186164683e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3147399902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4751342773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26370, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29963097870349886, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.12974539622664452, + "grad_norm": 0.0002280461776535958, + "kept_token_count": 573.2484375, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899516463279724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.925155360775534e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.71875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7697052001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26380, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.3004778355360031, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.13174477219581604, + "grad_norm": 0.00021852403006050736, + "kept_token_count": 572.165625, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906802237033844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.640564027591608e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.21864929199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7108581542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26390, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30040611028671266, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.12519670352339746, + "grad_norm": 0.00021326681599020958, + "kept_token_count": 572.571875, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904353618621826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.736208600457758e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.175537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3140014648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26400, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.30056376159191134, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.12061141654849053, + "grad_norm": 0.00022882693156134337, + "kept_token_count": 572.5546875, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9906201124191284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.664038267743308e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3683349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.23703918457034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.3003249496221542, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.12490426674485207, + "grad_norm": 0.00026015195180661976, + "kept_token_count": 572.521875, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902504026889801, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8084553307271565e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9666290283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8199523925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26420, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29936182498931885, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.13061846047639847, + "grad_norm": 0.00023156704264692962, + "kept_token_count": 573.234375, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900410830974579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.890218140440993e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.57763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.39100341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26430, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29942124485969546, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.12737897336483, + "grad_norm": 0.00021625557565130293, + "kept_token_count": 573.4390625, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911769926548004, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.446508326305775e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5781768798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3733337402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26440, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.29992879927158356, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.1241104319691658, + "grad_norm": 0.00017536987434141338, + "kept_token_count": 572.8890625, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910405397415161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.499809063214343e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.26091918945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9862091064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26450, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2993577241897583, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.12665844559669495, + "grad_norm": 0.00029881883529014885, + "kept_token_count": 573.371875, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904867291450501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.716142055054661e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6638977050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2499725341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26460, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2992193639278412, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.1239375725388527, + "grad_norm": 0.00028895019204355776, + "kept_token_count": 573.409375, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905718207359314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.682909737108275e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.01416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.44585571289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26470, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.2999622493982315, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.13155738562345504, + "grad_norm": 0.0002851937315426767, + "kept_token_count": 572.725, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904329478740692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7371524740592577e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.63681945800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1572509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26480, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2997038185596466, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.12952137589454651, + "grad_norm": 0.0002332425065105781, + "kept_token_count": 573.0953125, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990704494714737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6310850555310024e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6032775878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5423065185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26490, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2995541453361511, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.136664879322052, + "grad_norm": 0.0001888134138425812, + "kept_token_count": 573.021875, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905363321304321, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.696763451443985e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8153076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.99782409667966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26500, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.29945155382156374, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.13418198376893997, + "grad_norm": 0.0002442426048219204, + "kept_token_count": 573.328125, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907672047615051, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6065769018023273e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.39454956054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.95948181152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26510, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.300786617398262, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.11943653598427773, + "grad_norm": 0.00018819203251041472, + "kept_token_count": 572.3375, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911297678947448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.46495806297753e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9411346435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.92469177246096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26520, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.3006734371185303, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.1278220869600773, + "grad_norm": 0.0001991996105061844, + "kept_token_count": 572.0234375, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9910459220409393, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.497706020425539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6280456542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.12005310058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26530, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.3001342177391052, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.12542138770222663, + "grad_norm": 0.00021481553267221898, + "kept_token_count": 572.559375, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9900848746299744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8731174208805895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7130432128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5880401611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26540, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2998343676328659, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.12527658492326738, + "grad_norm": 0.00023736026196274906, + "kept_token_count": 572.9703125, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9904822528362274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.717895524459891e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3595489501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.26458129882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26550, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3001115292310715, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.12354129552841187, + "grad_norm": 0.00018440494022797793, + "kept_token_count": 572.7203125, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9902453720569611, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.810419366345741e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8729248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.66779174804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26560, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2997041493654251, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.14755321443080902, + "grad_norm": 0.00022560848447028548, + "kept_token_count": 572.971875, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905357956886292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.696968560689129e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.609130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5587585449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26570, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30023740530014037, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.1286587744951248, + "grad_norm": 0.0002967722248286009, + "kept_token_count": 572.5578125, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990865683555603, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.568116153473966e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2688385009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.25811462402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26580, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2995777606964111, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.12735603675246238, + "grad_norm": 0.0001874645531643182, + "kept_token_count": 573.065625, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990503066778183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.709763259394094e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9537078857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3630065917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26590, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.3004781395196915, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.13650713637471198, + "grad_norm": 0.00023394270101562142, + "kept_token_count": 572.328125, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9899284899234772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.934208434657194e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3937927246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9399475097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26600, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2993511974811554, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.13101001158356668, + "grad_norm": 0.00019145825353916734, + "kept_token_count": 573.2640625, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907992959022522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.594048030208796e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6171417236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5140686035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26610, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3003491222858429, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.12136334851384163, + "grad_norm": 0.0003584583755582571, + "kept_token_count": 572.465625, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9905037522315979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7094849903951396e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1389953613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8813201904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26620, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.300033301115036, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.13414923772215842, + "grad_norm": 0.00017238204600289464, + "kept_token_count": 572.7375, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0, + "mgd_cos_sim": 0.99053955078125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.695507366501261e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5232208251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.45042419433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26630, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2996900051832199, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.13398175090551376, + "grad_norm": 0.00021650099370162934, + "kept_token_count": 573.196875, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9911121428012848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.471844829618931e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.06397399902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3643432617188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26640, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.300426584482193, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.12034116014838218, + "grad_norm": 0.0002487344900146127, + "kept_token_count": 572.4453125, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0, + "mgd_cos_sim": 0.990249651670456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.808749243034981e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2622344970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1334503173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3018321394920349, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.1267651289701462, + "grad_norm": 0.0001936578337335959, + "kept_token_count": 571.2328125, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0, + "mgd_cos_sim": 0.9907397568225861, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.617300226324005e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9982421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3644744873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26660, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.29909762740135193, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.12246606051921845, + "grad_norm": 0.0002549842756707221, + "kept_token_count": 573.55625, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901682496070862, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.840546705760062e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.15723876953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2377868652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26670, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29957988262176516, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.1299851231276989, + "grad_norm": 0.00021967128850519657, + "kept_token_count": 573.128125, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906873285770417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.637781774159521e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.30372314453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.62318420410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26680, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2984627276659012, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.13024188876152037, + "grad_norm": 0.00024000037228688598, + "kept_token_count": 574.1765625, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905421435832977, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6944847306585873e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.63812255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5908538818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26690, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.30011217296123505, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.13637493103742598, + "grad_norm": 0.0001717085688142106, + "kept_token_count": 572.80625, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908131301403046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.588636282074731e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3249877929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.65712585449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26700, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.30038647055625917, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.12591471374034882, + "grad_norm": 0.00016701403365004808, + "kept_token_count": 572.6171875, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909153521060944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.548713320924435e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9259521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1947479248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26710, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.29979358017444613, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.11777513697743416, + "grad_norm": 0.000305837020277977, + "kept_token_count": 572.825, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902485132217407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.80919282179093e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9272064208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.59629516601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2992905259132385, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.13177327066659927, + "grad_norm": 0.00023872491146903485, + "kept_token_count": 573.325, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990255069732666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8066257184254936e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0411376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8689727783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26730, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29982490837574005, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.1270443394780159, + "grad_norm": 0.00020948713063262403, + "kept_token_count": 572.7375, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9897342681884765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.010066295450088e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0127288818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.06142272949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26740, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.2997962564229965, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.12647639140486716, + "grad_norm": 0.00019555706239771098, + "kept_token_count": 573.115625, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990416944026947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.743402521649841e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8732116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.36427612304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26750, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.2995049923658371, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.1268598511815071, + "grad_norm": 0.00017365708481520414, + "kept_token_count": 573.3296875, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902195155620575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.820523015747312e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.80560913085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4867828369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26760, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.30100257098674776, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.12815578132867814, + "grad_norm": 0.00017546732851769775, + "kept_token_count": 571.8609375, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909236192703247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.545474155544071e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.20924072265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.48179626464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26770, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.30014612674713137, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.12528892681002618, + "grad_norm": 0.0001587883016327396, + "kept_token_count": 572.81875, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903799414634704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.757858430617489e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.63551025390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3461853027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26780, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.29910862147808076, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.1333816632628441, + "grad_norm": 0.00033756718039512634, + "kept_token_count": 573.559375, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900247693061829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.896591697412077e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0936737060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1297607421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2988461434841156, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.13237696066498755, + "grad_norm": 0.0001886804966488853, + "kept_token_count": 573.703125, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904053926467895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.747912087419536e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3584716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.04026794433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26800, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29889858663082125, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.12532877251505853, + "grad_norm": 0.00016925593081396073, + "kept_token_count": 573.7078125, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905411541461945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.694876249937806e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2823791503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1040832519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.30001541376113894, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.12646913677453994, + "grad_norm": 0.0003681414236780256, + "kept_token_count": 572.715625, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990167886018753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.840678800770547e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8362548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01470642089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26820, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2988863945007324, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.1262126699090004, + "grad_norm": 0.00024130735255312175, + "kept_token_count": 573.8484375, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902211129665375, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.81989506422542e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.778515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.37603149414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26830, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.29987852573394774, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.11709465533494949, + "grad_norm": 0.0001919099740916863, + "kept_token_count": 572.8828125, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908525466918945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.573235953808762e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9169067382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9051788330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26840, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2993078172206879, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.13417448848485947, + "grad_norm": 0.00020039569062646478, + "kept_token_count": 573.409375, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9896454632282257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.0447583523928186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8637969970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3020477294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26850, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29972673654556276, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.13531131744384767, + "grad_norm": 0.00026176971732638776, + "kept_token_count": 572.975, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905144691467285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7053033884149046e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.19037475585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.49707946777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29891242682933805, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.12385386750102043, + "grad_norm": 0.0002272474521305412, + "kept_token_count": 573.6140625, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905843794345855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.677997083286755e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.46656494140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8781494140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26870, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30090445280075073, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.1253798484802246, + "grad_norm": 0.0001744372711982578, + "kept_token_count": 572.259375, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909852206707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.521415474097012e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4485290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.00200805664065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26880, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30072948932647703, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.12920714542269707, + "grad_norm": 0.00021730548178311437, + "kept_token_count": 572.090625, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904785752296448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7193232128629464e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2309631347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1344421386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26890, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29965285658836366, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.12441371604800225, + "grad_norm": 0.00015517286374233663, + "kept_token_count": 573.075, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906890749931335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.63710118108429e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.793701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.90220336914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26900, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2996786653995514, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.12440533265471458, + "grad_norm": 0.00029933537007309496, + "kept_token_count": 573.103125, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902143239974975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.822542021225672e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0147247314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.46751098632814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26910, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29983603954315186, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.1326930396258831, + "grad_norm": 0.0001916070468723774, + "kept_token_count": 572.890625, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990665203332901, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.646426775958389e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.54766235351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8951446533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26920, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3001189947128296, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.13379051759839058, + "grad_norm": 0.00028339007985778153, + "kept_token_count": 572.6625, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990798830986023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.594219797378173e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2003662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.69532165527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26930, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30002348124980927, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.12254233434796333, + "grad_norm": 0.0001922038063639775, + "kept_token_count": 572.7890625, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908776164054871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.563451573427301e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8298675537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8580047607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26940, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.30000083446502684, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.12496404200792313, + "grad_norm": 0.00021934905089437962, + "kept_token_count": 572.753125, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9910815834999085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.483772106847027e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2052032470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0172637939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26950, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.30046945810317993, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.12935585603117944, + "grad_norm": 0.00028528441907837987, + "kept_token_count": 572.3375, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901725471019744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.83886435884051e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4629669189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6416778564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26960, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.299624028801918, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.12845849320292474, + "grad_norm": 0.00023053886252455413, + "kept_token_count": 572.959375, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909423649311065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.538157616276294e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.22767639160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.55059509277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26970, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30017273128032684, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.1311303660273552, + "grad_norm": 0.0002796496555674821, + "kept_token_count": 572.6765625, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907215714454651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6244064904167314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0049652099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0270233154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26980, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30020962059497835, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.12584223374724388, + "grad_norm": 0.00020841986406594515, + "kept_token_count": 572.490625, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903956353664398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7517257806030105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0638793945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6564971923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26990, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.30063540637493136, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.1305997408926487, + "grad_norm": 0.00029629291384480894, + "kept_token_count": 572.38125, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906103730201721, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6678457217931284e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.20212707519534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0094299316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27000, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.29906401336193084, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.12981643974781037, + "grad_norm": 0.00013113749446347356, + "kept_token_count": 573.321875, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9913113832473754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.394010127522051e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.82490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70945739746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27010, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.30068058967590333, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.12841300442814826, + "grad_norm": 0.0002489805337972939, + "kept_token_count": 572.221875, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909571945667267, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.532365954015404e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7117034912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3037963867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27020, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29922378063201904, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.12807337790727616, + "grad_norm": 0.0002349477435927838, + "kept_token_count": 573.4875, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906194329261779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6642979102907704e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.32500610351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6412658691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27030, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.300328665971756, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.12902109548449517, + "grad_norm": 0.0002522720897104591, + "kept_token_count": 572.7265625, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990590637922287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.675552397908177e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.78672790527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2345458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27040, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3003692597150803, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.12324020341038704, + "grad_norm": 0.0002365267282584682, + "kept_token_count": 572.7125, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906140685081481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.66639913409017e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.63788146972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.28416137695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27050, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.30002276599407196, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.12925463542342186, + "grad_norm": 0.00019737370894290507, + "kept_token_count": 573.0, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908876717090607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.559525321179535e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6149078369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.27814331054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27060, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.3014128029346466, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.1350131079554558, + "grad_norm": 0.0002458937233313918, + "kept_token_count": 571.6125, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9898603320121765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9608242877875456e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.54892883300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7102020263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27070, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2996126711368561, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.13673469573259353, + "grad_norm": 0.00018912529048975557, + "kept_token_count": 573.125, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990304458141327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.787339701375458e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.47247314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0928009033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27080, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.29979275166988373, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.12527222856879233, + "grad_norm": 0.00028868942172266543, + "kept_token_count": 572.9671875, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907418012619018, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.616507565311622e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.91437377929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7662414550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.2995288997888565, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.12465033158659936, + "grad_norm": 0.00018528546206653118, + "kept_token_count": 573.0375, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904549539089202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7285478538251485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.15618591308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4058349609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27100, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.2990473836660385, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.12568895518779755, + "grad_norm": 0.00021608627866953611, + "kept_token_count": 573.6125, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990777313709259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6026308589498515e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.04185791015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36705322265624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3004262626171112, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.11800029799342156, + "grad_norm": 0.00019543942471500486, + "kept_token_count": 572.43125, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907596111297607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.609545437939232e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5301513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.33057861328126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27120, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2994274437427521, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.12427999526262283, + "grad_norm": 0.00015794936916790903, + "kept_token_count": 573.559375, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904683470726013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.723319423443172e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4165069580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.71320190429685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27130, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.29998567402362825, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.12526659443974494, + "grad_norm": 0.00017788997502066195, + "kept_token_count": 572.828125, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902911424636841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.792542338487692e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0293212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6267364501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27140, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.30045930445194247, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.11855415552854538, + "grad_norm": 0.0001646494201850146, + "kept_token_count": 572.2734375, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905863344669342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6772227758774534e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4408294677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7732177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27150, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.3003131031990051, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.11797538697719574, + "grad_norm": 0.0002395763440290466, + "kept_token_count": 572.615625, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904024541378021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7490593967959285e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6600616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.39564514160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27160, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2989526063203812, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.12976518720388414, + "grad_norm": 0.00019497230823617429, + "kept_token_count": 573.8, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909199595451355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5469142676447515e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.93128662109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.73862915039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27170, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.29991450905799866, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.12348025813698768, + "grad_norm": 0.00019270970369689167, + "kept_token_count": 572.7453125, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902951180934906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.790990485867951e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4034790039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0297576904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27180, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.2988508760929108, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.12491218000650406, + "grad_norm": 0.00027953123208135366, + "kept_token_count": 573.8234375, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904573082923889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7276369766914286e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.70354919433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.18878479003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27190, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3000887542963028, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.13686092868447303, + "grad_norm": 0.0002091702917823568, + "kept_token_count": 572.859375, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904844701290131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7170170253375544e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.60570068359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6537780761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27200, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.30010123252868653, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.1260994590818882, + "grad_norm": 0.0002867900475393981, + "kept_token_count": 572.853125, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990234500169754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.814666015387047e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1611022949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.10087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27210, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.29966185092926023, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.1174800530076027, + "grad_norm": 0.00031981003121472895, + "kept_token_count": 573.059375, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909402191638946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.538991149980575e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1980285644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.88872985839845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2992444097995758, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.12886095941066741, + "grad_norm": 0.00022664139396511018, + "kept_token_count": 573.4671875, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904778122901916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.719624291989021e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.40526428222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7752288818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27230, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29976700246334076, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.13177422434091568, + "grad_norm": 0.00020778871839866042, + "kept_token_count": 572.965625, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907115876674653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.628302947618067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4876007080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4629272460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27240, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29993427395820615, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.1321417808532715, + "grad_norm": 0.00016987597336992621, + "kept_token_count": 572.7984375, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909701645374298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5273010144010183e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.17534790039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.55455017089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30030952095985414, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.12978337183594704, + "grad_norm": 0.00019980977231170982, + "kept_token_count": 572.49375, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905333340167999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6979352444177493e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9861175537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6684844970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27260, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2997690439224243, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.12218675762414932, + "grad_norm": 0.0001677481341175735, + "kept_token_count": 572.9546875, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909041345119476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5530908644432205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3276702880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3459991455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27270, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30037218928337095, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.1323034830391407, + "grad_norm": 0.00017758052854333073, + "kept_token_count": 572.5625, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904878675937653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.715697439474752e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8196166992187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.94049377441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27280, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3007215827703476, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.13005056455731392, + "grad_norm": 0.00032395662856288254, + "kept_token_count": 571.965625, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907006978988647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6325563087302726e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3576599121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2499481201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27290, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.29969675838947296, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.12273804992437362, + "grad_norm": 0.00022561637160833925, + "kept_token_count": 573.009375, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905249536037445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7012128814239986e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4308837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8787841796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27300, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2992223560810089, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.13731892704963683, + "grad_norm": 0.0001786938519217074, + "kept_token_count": 573.484375, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909010827541351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.554281502147205e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.42845153808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.02385864257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27310, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2988647848367691, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.12895507588982583, + "grad_norm": 0.00018347245350014418, + "kept_token_count": 573.83125, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990848046541214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.575007285689935e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6426208496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4784790039063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27320, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.3010948747396469, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.12861424759030343, + "grad_norm": 0.00020297813171055168, + "kept_token_count": 571.7390625, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0, + "mgd_cos_sim": 0.991483348608017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.326831429149024e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7435821533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24334106445315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27330, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.29896520972251894, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.13374239578843117, + "grad_norm": 0.00013953723828308284, + "kept_token_count": 573.6, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901040434837342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.865623621095438e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.79779052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0541107177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27340, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.299350705742836, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.125265996158123, + "grad_norm": 0.0001734107790980488, + "kept_token_count": 573.2875, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990863972902298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.568775282474235e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9356170654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4608673095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27350, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29955623745918275, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.12374328151345253, + "grad_norm": 0.00022735378297511488, + "kept_token_count": 573.35625, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905374825000763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.696314561238978e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8027374267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0065155029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27360, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.2997982144355774, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.12207998409867286, + "grad_norm": 0.0002572893863543868, + "kept_token_count": 572.946875, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905341744422913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6976073897676545e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.75034790039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.48256530761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27370, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3003725647926331, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.12677315324544908, + "grad_norm": 0.0002442107943352312, + "kept_token_count": 572.3640625, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902253091335297, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.818259938270785e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6344482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.34130859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27380, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.29909173846244813, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.13074346929788588, + "grad_norm": 0.0002477586385793984, + "kept_token_count": 573.4375, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903426170349121, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7724331559729765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2327911376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.94629821777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27390, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.3011419862508774, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.12859323099255562, + "grad_norm": 0.00021917169215157628, + "kept_token_count": 571.85625, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906004548072815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.671712802315597e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1306182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0771209716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.29936074912548066, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.12425936385989189, + "grad_norm": 0.00022873586567584425, + "kept_token_count": 573.20625, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900572538375855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8839019543956967e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.01670532226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9864105224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27410, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3008966535329819, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.12575097903609275, + "grad_norm": 0.00019436859292909503, + "kept_token_count": 571.7453125, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906910181045532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6363386243465357e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4350921630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9946563720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27420, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.3012039691209793, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.1356395162642002, + "grad_norm": 0.00017913254851009697, + "kept_token_count": 571.7234375, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908980607986451, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5554691021388865e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.993994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17615966796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27430, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.2998394101858139, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.12840477079153062, + "grad_norm": 0.0001981501409318298, + "kept_token_count": 572.8984375, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909942150115967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5179009137209506e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2715362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0907897949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27440, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2991949528455734, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.13661779314279557, + "grad_norm": 0.00020703197515103966, + "kept_token_count": 573.4671875, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906693458557129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.644806965894531e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.411669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.70733947753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27450, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30021447837352755, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.12849766090512277, + "grad_norm": 0.00021592002303805202, + "kept_token_count": 572.671875, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906438112258911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6547835043165834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4469299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8088653564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27460, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3002863675355911, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.12960590794682503, + "grad_norm": 0.00027113239048048854, + "kept_token_count": 572.4859375, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907549798488617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.611353549786145e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.67040100097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8067687988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27470, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29984436333179476, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.13555072024464607, + "grad_norm": 0.00020719383610412478, + "kept_token_count": 573.025, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906690835952758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.644903699751012e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2392150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1747161865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27480, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.300495046377182, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.13056433200836182, + "grad_norm": 0.00018370129691902548, + "kept_token_count": 572.2015625, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9918117523193359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.198559970769565e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.80621643066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.96689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27490, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2996464937925339, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.12362722903490067, + "grad_norm": 0.0002290170668857172, + "kept_token_count": 573.175, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907959878444672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.59533769369591e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5350677490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.40794067382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27500, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29959276020526887, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.12861328423023224, + "grad_norm": 0.00024847930762916803, + "kept_token_count": 572.8921875, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901974976062775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.829119996225927e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.08649291992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2361663818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27510, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30025562047958376, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.13366029635071755, + "grad_norm": 0.00017850034055300057, + "kept_token_count": 572.665625, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906319797039032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6593993718270214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6357055664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6182403564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27520, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.29975556433200834, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.12348478883504868, + "grad_norm": 0.00022890776745043695, + "kept_token_count": 573.1640625, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902885377407074, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.793556643358897e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.91266479492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.13816223144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27530, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.3008397459983826, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.1283906728029251, + "grad_norm": 0.00019267378957010806, + "kept_token_count": 572.3140625, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908279299736023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.582853914849693e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4132476806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5672607421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27540, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.29958377182483675, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.12161863669753074, + "grad_norm": 0.00020847213454544544, + "kept_token_count": 573.1171875, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904197096824646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7423213143483736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3220153808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5195770263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27550, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.29926642775535583, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.1303180582821369, + "grad_norm": 0.00021463067969307303, + "kept_token_count": 573.3546875, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906144499778747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.666256270662416e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.85589904785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.11139526367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27560, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29962384700775146, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.12781909629702567, + "grad_norm": 0.0002640046295709908, + "kept_token_count": 573.3046875, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907994449138642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.593981946323765e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7516235351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3660827636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27570, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.3001526713371277, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.12196485623717308, + "grad_norm": 0.00021673974697478116, + "kept_token_count": 572.8203125, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904966413974762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7122694084246175e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.28468322753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0790618896484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27580, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.3001347452402115, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.11721605956554412, + "grad_norm": 0.00017030337767209858, + "kept_token_count": 572.575, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904185056686401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.74278926756233e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1514953613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.52842407226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27590, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.300054544210434, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.12063363566994667, + "grad_norm": 0.00021196345915086567, + "kept_token_count": 572.925, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909860134124756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.521101716614794e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4108184814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.74338989257814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27600, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.3003836005926132, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.1267000198364258, + "grad_norm": 0.0002290226984769106, + "kept_token_count": 572.4328125, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909673869609833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5283824036014264e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.45830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8194061279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27610, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.29897820949554443, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.1269890159368515, + "grad_norm": 0.00017538185056764632, + "kept_token_count": 573.85625, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907431185245514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.615989444369916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8014678955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0352020263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27620, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.3002911448478699, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.13627461865544319, + "grad_norm": 0.0002369398280279711, + "kept_token_count": 572.70625, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903775751590729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.758776329050306e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.29607849121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01219177246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27630, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.30135208666324614, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.13118340969085693, + "grad_norm": 0.00016408012015745044, + "kept_token_count": 571.7578125, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904723346233368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.721764478541445e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1428466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8602752685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27640, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29967852830886843, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.12598642259836196, + "grad_norm": 0.00021589263633359224, + "kept_token_count": 573.0390625, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990571242570877, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.68313085346017e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5114288330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.67283325195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27650, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3000170350074768, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.1315930426120758, + "grad_norm": 0.0002362647937843576, + "kept_token_count": 572.9625, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990429139137268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.738642335520126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3674285888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9025482177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27660, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.2990271389484406, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.12679685950279235, + "grad_norm": 0.00036657185410149395, + "kept_token_count": 573.496875, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990313982963562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.783615684369579e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5588104248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1524322509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27670, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2994623690843582, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.13243810310959817, + "grad_norm": 0.00022481738415081054, + "kept_token_count": 573.0875, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909830391407013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.522273764247075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1107116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3179138183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27680, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.300287264585495, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.12744006365537644, + "grad_norm": 0.00019591710588429123, + "kept_token_count": 572.6640625, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901406109333039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8513368053827435e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.59466857910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6886352539062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27690, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30030240416526793, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.12156759649515152, + "grad_norm": 0.00017793850565794855, + "kept_token_count": 572.2578125, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908279657363892, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5828460931952576e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0985473632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3100341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27700, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.3004764914512634, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.1303844690322876, + "grad_norm": 0.00020809903799090534, + "kept_token_count": 572.190625, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990219509601593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.820528690994252e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9200378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.57944946289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27710, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.30097076296806335, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.14158000573515891, + "grad_norm": 0.0002067078312393278, + "kept_token_count": 572.0296875, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903046071529389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.787280038523022e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6210601806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.01429138183596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30103552937507627, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.13715074881911277, + "grad_norm": 0.0002621801395434886, + "kept_token_count": 571.7390625, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902313113212585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8159118048497476e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.25982666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.41028747558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27730, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.3005130857229233, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.12944869101047515, + "grad_norm": 0.0002090642519760877, + "kept_token_count": 572.2625, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908767104148865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5638044937513766e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.05638122558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.27412109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.3012166142463684, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.12420182451605796, + "grad_norm": 0.0002723754441831261, + "kept_token_count": 571.6625, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990321671962738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7806184263899925e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6436492919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.31292114257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27750, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3001197069883347, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.1208728313446045, + "grad_norm": 0.00023080006940290332, + "kept_token_count": 572.7890625, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905193328857422, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.703408256114926e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.09747314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1087982177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27760, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.2997613251209259, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.11961817517876625, + "grad_norm": 0.0002416841743979603, + "kept_token_count": 573.0375, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905727088451386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.682555179693736e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8701477050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.483349609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27770, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.299739345908165, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.13277072384953498, + "grad_norm": 0.00029871740844100714, + "kept_token_count": 572.9890625, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990736311674118, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.618645569076761e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0371917724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.83775634765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27780, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3001652806997299, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.12686330527067186, + "grad_norm": 0.0002842071990016848, + "kept_token_count": 572.7640625, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904717326164245, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.721997836692026e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0343444824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.11168212890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27790, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2994355529546738, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.1295593000948429, + "grad_norm": 0.0001985910057555884, + "kept_token_count": 573.1828125, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990619444847107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6642908889916724e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4051788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.46861572265624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27800, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.3007974177598953, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.13658232018351554, + "grad_norm": 0.00018924972391687334, + "kept_token_count": 572.0390625, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903776347637177, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7587535916827616e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.23069763183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3186279296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27810, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29952948093414306, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.13113616779446602, + "grad_norm": 0.00024177489103749394, + "kept_token_count": 573.15, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908677041530609, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.56732532964088e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2679107666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4432739257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27820, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2999179780483246, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.1320832885801792, + "grad_norm": 0.0001994200429180637, + "kept_token_count": 572.8390625, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9896320700645447, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.04999667807715e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5885040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4463256835937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27830, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3001765549182892, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.11862079277634621, + "grad_norm": 0.0002778339257929474, + "kept_token_count": 572.565625, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9895962595939636, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.06398088671267e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0613647460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0354919433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27840, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.3006424009799957, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.12988494858145713, + "grad_norm": 0.0001845063961809501, + "kept_token_count": 572.325, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904352486133575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.736246799235232e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.935400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.18409729003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30008178055286405, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.12341072633862496, + "grad_norm": 0.00022595205518882722, + "kept_token_count": 572.903125, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904999494552612, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.710976016009226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1857391357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.92743225097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27860, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30030318200588224, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.12235766872763634, + "grad_norm": 0.00017085052968468517, + "kept_token_count": 572.471875, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990114688873291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8614708682871426e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2980712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1420837402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27870, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30033718049526215, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.13280144557356835, + "grad_norm": 0.00021283024398144335, + "kept_token_count": 572.4546875, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909118115901947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.550094479578547e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2254180908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.88844299316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27880, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2998742163181305, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.13009897917509078, + "grad_norm": 0.00023781826894264668, + "kept_token_count": 572.621875, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901633739471436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.842449168587336e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2849548339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1972961425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27890, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.2998394161462784, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.13089851662516594, + "grad_norm": 0.0002953563816845417, + "kept_token_count": 572.95625, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906665682792664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6458919203141706e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3009918212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.82074584960935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27900, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30036401748657227, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.1353088915348053, + "grad_norm": 0.00020598864648491144, + "kept_token_count": 572.3140625, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909152746200561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.54873849573778e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.01414489746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.79351196289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27910, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.30026814341545105, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.1350986398756504, + "grad_norm": 0.00021087346249260008, + "kept_token_count": 572.409375, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907353937625885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.619009694375563e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5220611572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2287841796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27920, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.29987787902355195, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.12282214611768723, + "grad_norm": 0.0002857111394405365, + "kept_token_count": 572.8734375, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900658965110779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.880527983710635e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.72674560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2274963378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27930, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.300125515460968, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.12169173136353492, + "grad_norm": 0.00020944721472915262, + "kept_token_count": 572.69375, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907935917377472, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.59626723366091e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3832733154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7135070800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.3005682319402695, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.13163793832063675, + "grad_norm": 0.00025610969169065356, + "kept_token_count": 572.2046875, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990940374135971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5389347249292766e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.17864990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8771514892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27950, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29973492324352263, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.13521699011325836, + "grad_norm": 0.00024109920195769519, + "kept_token_count": 572.896875, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906466603279114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.653665517049376e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5523223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2390838623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27960, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29962087869644166, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.13480169698596, + "grad_norm": 0.00020259743905626237, + "kept_token_count": 573.1, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9897171378135681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.016760140075348e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.32188110351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7543670654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2996852666139603, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.12290936484932899, + "grad_norm": 0.0002415881463093683, + "kept_token_count": 572.8734375, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990721321105957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.624505625339225e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3007568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.72190246582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27980, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.2993714302778244, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.1307343438267708, + "grad_norm": 0.00018455238023307174, + "kept_token_count": 573.2890625, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900952756404877, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.869049869535957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.32207946777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.25466003417966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27990, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2988790988922119, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.12400815561413765, + "grad_norm": 0.0003949386882595718, + "kept_token_count": 573.765625, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900932133197784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.869856445817277e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.97666931152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2481658935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28000, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29983527660369874, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.13395441472530364, + "grad_norm": 0.00022675309446640313, + "kept_token_count": 572.921875, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990859431028366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5705594564205966e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.41710510253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.16778259277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28010, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29962230920791627, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.12977494522929192, + "grad_norm": 0.00019567727576941252, + "kept_token_count": 573.0921875, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9910391092300415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5003678567591125e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9577331542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9372131347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30012843310832976, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.12993372231721878, + "grad_norm": 0.00021990700042806566, + "kept_token_count": 572.7296875, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902520477771759, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.807813191087916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6665405273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.58132019042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28030, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2997170090675354, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.13217251524329185, + "grad_norm": 0.0002273658465128392, + "kept_token_count": 572.96875, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9912678241729737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.411024226807058e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.71900634765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.96412658691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28040, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.3008782535791397, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.12756930217146872, + "grad_norm": 0.0002288116083946079, + "kept_token_count": 572.1234375, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907764613628387, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6029604234499854e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.563232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36326904296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28050, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3007157504558563, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.13373298570513725, + "grad_norm": 0.00017342198407277465, + "kept_token_count": 572.190625, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990890520811081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5584085890150166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.18829650878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3356506347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28060, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30056825280189514, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.11820564642548562, + "grad_norm": 0.0002867647272069007, + "kept_token_count": 572.440625, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906240940093994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.662482158688363e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6020446777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.64854736328124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28070, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30024493932724, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.1418587125837803, + "grad_norm": 0.00020980180124752223, + "kept_token_count": 572.4203125, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904225826263428, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.741194632311817e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.24729309082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4160583496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28080, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.29989500641822814, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.13271537721157073, + "grad_norm": 0.00031029016827233136, + "kept_token_count": 572.8609375, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909765303134919, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.524814947013511e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.51755676269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.86671142578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28090, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3008049577474594, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.12384501621127128, + "grad_norm": 0.00018328151782043278, + "kept_token_count": 572.1234375, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905208826065064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7028018778073604e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1834197998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6615966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28100, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30022063851356506, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.12974731624126434, + "grad_norm": 0.0002641726459842175, + "kept_token_count": 572.5953125, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9899229288101197, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9363744872389364e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4405975341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6020812988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28110, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.30028857588768004, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.12709004208445548, + "grad_norm": 0.0002187992649851367, + "kept_token_count": 572.478125, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908358633518219, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.579763615562115e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8844787597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.97528076171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28120, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29949812591075897, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.13625651895999907, + "grad_norm": 0.00020260881865397096, + "kept_token_count": 573.3609375, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909327924251556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.541894693626091e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9068603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.37018127441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28130, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29884942471981046, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.12312334403395653, + "grad_norm": 0.00014660583110526204, + "kept_token_count": 573.8046875, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909401357173919, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.539029239618685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7511413574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.67822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.29964507818222047, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.1176178701221943, + "grad_norm": 0.00020639208378270268, + "kept_token_count": 573.0875, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902387201786041, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.813018993241713e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.06336364746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4178741455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28150, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3009885400533676, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.13406105935573578, + "grad_norm": 0.00017549641779623926, + "kept_token_count": 572.0796875, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905489981174469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6918148362019564e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.38584594726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.26806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28160, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3001433849334717, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.12978926971554755, + "grad_norm": 0.00021441892022266984, + "kept_token_count": 572.6296875, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902009248733521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.827785549219698e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5031799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6735778808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28170, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3000029921531677, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.1287914976477623, + "grad_norm": 0.00025906856171786785, + "kept_token_count": 572.890625, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906582593917846, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.649140198831446e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.87872619628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.51552429199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.30074697732925415, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.1326463468372822, + "grad_norm": 0.00020567546016536653, + "kept_token_count": 571.96875, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9912392675876618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.422182107897243e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4920257568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5180969238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28190, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.29964075088500974, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.12467074319720269, + "grad_norm": 0.0002013191260630265, + "kept_token_count": 573.0125, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907987833023071, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5942505928687754e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.00384216308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2362823486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2993317633867264, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.12106162309646606, + "grad_norm": 0.00018387599266134202, + "kept_token_count": 573.3890625, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907412886619568, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6167038160783704e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0919494628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82029724121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28210, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3007989227771759, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.11765499711036682, + "grad_norm": 0.00021867141185794026, + "kept_token_count": 572.50625, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9910615444183349, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4916048389277424e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.15105590820315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2069091796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28220, + "valid_token_count": 818.796875 + }, + { + "actual_token_mask_ratio": 0.2996168315410614, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.12048996686935425, + "grad_norm": 0.00021476724941749126, + "kept_token_count": 573.134375, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908536434173584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.572805726435035e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.91893310546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6834655761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28230, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.3000447303056717, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.1286010652780533, + "grad_norm": 0.0001926361583173275, + "kept_token_count": 572.8375, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906544864177704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.650613653007895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.83998107910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.89360961914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28240, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2994448184967041, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.12919836714863778, + "grad_norm": 0.00023081251129042357, + "kept_token_count": 573.1578125, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906698703765869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6446065496420485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.94695739746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3602233886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28250, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30036875903606414, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.1263594798743725, + "grad_norm": 0.00017066513828467578, + "kept_token_count": 572.403125, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990890109539032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5585668956628067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2101226806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4283782958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28260, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2996326744556427, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.12871915623545646, + "grad_norm": 0.00021218310575932264, + "kept_token_count": 573.1453125, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906496644020081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6524896859191355e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9917358398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2450836181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28270, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.29942815005779266, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.12801729664206504, + "grad_norm": 0.0001977149659069255, + "kept_token_count": 573.03125, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902799546718597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.796911078097764e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.74896240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4969421386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28280, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.29916190803050996, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.126135466247797, + "grad_norm": 0.00018895059474743903, + "kept_token_count": 573.4640625, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908386468887329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.578675277822185e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6089569091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.20783081054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28290, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2996958702802658, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.1274846911430359, + "grad_norm": 0.00021622813073918223, + "kept_token_count": 572.9671875, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901349246501923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.853561502182856e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4405181884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2656219482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28300, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3007515728473663, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.1397468328475952, + "grad_norm": 0.00027244206285104156, + "kept_token_count": 572.2890625, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906559765338898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6500317946774885e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0300018310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.11536865234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28310, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2994028329849243, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.1232388123869896, + "grad_norm": 0.00021951286180410534, + "kept_token_count": 572.9921875, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905717492103576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.682933856907766e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.76673889160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.23032531738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28320, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.29929725229740145, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.12747454568743705, + "grad_norm": 0.00021695805480703712, + "kept_token_count": 573.2890625, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903344213962555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.775632721954025e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.12700805664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1660949707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28330, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.30110991895198824, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.12452156096696854, + "grad_norm": 0.0002521613205317408, + "kept_token_count": 571.940625, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9899253368377685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.935431923309807e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3675537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4412872314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28340, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30087587237358093, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.11851419359445572, + "grad_norm": 0.0001714812096906826, + "kept_token_count": 572.259375, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905333697795868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.69792749552289e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8400360107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8446899414063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28350, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.29948264062404634, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.12693928107619284, + "grad_norm": 0.0002395344927208498, + "kept_token_count": 573.04375, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990203857421875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.826639258477371e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7469787597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5527862548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28360, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.29996891915798185, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.12805866673588753, + "grad_norm": 0.00025995788746513426, + "kept_token_count": 572.6265625, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9897205173969269, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.01544170017587e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.25706176757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.48618774414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28370, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.2994227796792984, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.12886077761650086, + "grad_norm": 0.00030855947989039123, + "kept_token_count": 573.078125, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904053986072541, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7479117963812314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8990844726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4105499267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28380, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.29952951371669767, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.12721953988075257, + "grad_norm": 0.00021143784397281706, + "kept_token_count": 573.1671875, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9916267037391663, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.270840697950916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7848266601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.2652282714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28390, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2992250472307205, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.11980918720364571, + "grad_norm": 0.0003066735516767949, + "kept_token_count": 573.5234375, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901986122131348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8286811832222154e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.81708374023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4553283691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28400, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.2999661505222321, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.13145456239581108, + "grad_norm": 0.00017192804079968482, + "kept_token_count": 572.8890625, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904151260852814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7441131644300186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3590423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.45741882324216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28410, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.2999521791934967, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.12389397397637367, + "grad_norm": 0.00019937306933570653, + "kept_token_count": 573.0265625, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0, + "mgd_cos_sim": 0.991141813993454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4602515006554314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7349060058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.80095825195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28420, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.3001758724451065, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.127469689399004, + "grad_norm": 0.00022909039398655295, + "kept_token_count": 572.703125, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907506346702576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6130474472884086e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.06444396972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.35728149414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.299534672498703, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.12742753103375434, + "grad_norm": 0.00036616832949221134, + "kept_token_count": 573.3, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903398513793945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.773514908971265e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8716735839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.94086303710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28440, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.2994278401136398, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.12663234099745752, + "grad_norm": 0.0002101505087921396, + "kept_token_count": 573.3828125, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909301817417144, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.542918439052301e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4717620849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.45537719726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.30014200806617736, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.12042451277375221, + "grad_norm": 0.00025197784998454154, + "kept_token_count": 572.853125, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904034912586213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.748650087800343e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2699401855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6692138671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28460, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.30132982730865476, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.13418470770120622, + "grad_norm": 0.00022108123812358826, + "kept_token_count": 571.540625, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905124425888061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.706091592903249e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9502380371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.87919921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28470, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.3004259943962097, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.12989297956228257, + "grad_norm": 0.00019672811322379857, + "kept_token_count": 572.4953125, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906225740909577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.663074676296674e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.176513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6005798339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29915049076080324, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.12634814977645875, + "grad_norm": 0.0002591370139271021, + "kept_token_count": 573.4109375, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901407599449158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8512911123689265e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9676025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17852783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28490, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2991645514965057, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.13089072778820993, + "grad_norm": 0.00015972720575518906, + "kept_token_count": 573.5734375, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9898791670799255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9534696043119766e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.90911865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8146118164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28500, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2988841116428375, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.12471989914774895, + "grad_norm": 0.00020438309002202004, + "kept_token_count": 573.575, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904622197151184, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7257176882121715e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.25295104980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6774505615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28510, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.29927178621292116, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.12596577629446984, + "grad_norm": 0.00020343020150903612, + "kept_token_count": 573.55625, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907450079917908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.615248097048607e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4146240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.73435974121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28520, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.2998259663581848, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.11891102939844131, + "grad_norm": 0.00018237772746942937, + "kept_token_count": 572.871875, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9911633312702179, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.451841439527925e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3798797607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28530, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3002770334482193, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.13229529783129693, + "grad_norm": 0.00018491903028916568, + "kept_token_count": 572.6390625, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9912646830081939, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.412253499845974e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.18788757324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.87229309082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28540, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.30073711574077605, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.11930201351642608, + "grad_norm": 0.00021271493460517377, + "kept_token_count": 572.071875, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903721988201142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.760878207685892e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.68410339355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9454772949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28550, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.29996253848075866, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.12096230089664459, + "grad_norm": 0.0001919119677040726, + "kept_token_count": 572.9734375, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0, + "mgd_cos_sim": 0.991175377368927, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4471389517420904e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.51690063476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9739166259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28560, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.30035039186477663, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.12642526179552077, + "grad_norm": 0.00026869482826441526, + "kept_token_count": 572.5859375, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990522038936615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7023481490905395e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1494598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5704864501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28570, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.29982961118221285, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.12829882875084878, + "grad_norm": 0.0002139933785656467, + "kept_token_count": 572.9015625, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907256841659546, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.622805161285214e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.87391967773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3398681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28580, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.30009545385837555, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.12848699763417243, + "grad_norm": 0.0002121641591656953, + "kept_token_count": 572.70625, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906023681163788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.670966470963322e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6253967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.08550720214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28590, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3004401296377182, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.12488648071885108, + "grad_norm": 0.000204959258553572, + "kept_token_count": 572.6421875, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904337525367737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7368343691923654e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5442810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0270721435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28600, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.2998829185962677, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.12812102660536767, + "grad_norm": 0.0003151060955133289, + "kept_token_count": 573.0203125, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909616947174072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.530608446453698e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6961608886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6748352050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.30016019344329836, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.12722139209508895, + "grad_norm": 0.00027475692331790924, + "kept_token_count": 572.6984375, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990374630689621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.759926148632076e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.78779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.77117919921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28620, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2990534991025925, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.13034274950623512, + "grad_norm": 0.00018178544996771961, + "kept_token_count": 573.6375, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9910451769828796, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4979988595296166e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.12464599609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.75465698242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28630, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.3010985881090164, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.1288502797484398, + "grad_norm": 0.00028656848007813096, + "kept_token_count": 571.93125, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900406539440155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.890384796250146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5058227539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.90003967285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28640, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30076209306716917, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.12287705540657043, + "grad_norm": 0.00030527892522513866, + "kept_token_count": 572.0203125, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906800806522369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.640610048023518e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.78876037597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5826934814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28650, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2997643440961838, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.13156764581799507, + "grad_norm": 0.00025591623852960765, + "kept_token_count": 572.7125, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9899718105793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.917286485375371e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6936920166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.02651062011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28660, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.3003969967365265, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.128192301094532, + "grad_norm": 0.00018899184942711145, + "kept_token_count": 572.2765625, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9912550747394562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.415999817661941e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99737243652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.28041076660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28670, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.3005718946456909, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.12391119226813316, + "grad_norm": 0.0003147199749946594, + "kept_token_count": 572.45, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908765196800232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.563875361578539e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.95754089355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7579345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28680, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.3002817064523697, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.1322343908250332, + "grad_norm": 0.00020818128541577607, + "kept_token_count": 572.590625, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905671954154969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.684711518872064e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9648376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8794281005859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28690, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.30116316974163054, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.12832133322954178, + "grad_norm": 0.00023841603251639754, + "kept_token_count": 571.9671875, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907630980014801, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608188817452174e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1419616699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2246551513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28700, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.29972130358219146, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.12566545829176903, + "grad_norm": 0.00014717807061970234, + "kept_token_count": 572.934375, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907458186149597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.614937268139329e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4710754394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9847900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28710, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2984287887811661, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.128835965692997, + "grad_norm": 0.00017211669182870537, + "kept_token_count": 574.084375, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9900541543960572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8851167118991725e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5676788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.82366638183595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28720, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3006999582052231, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.12869387045502662, + "grad_norm": 0.000154380380990915, + "kept_token_count": 572.5078125, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990854275226593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5725686029763895e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.52704162597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8229644775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28730, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.30104528069496156, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.12551817893981934, + "grad_norm": 0.00022896393784321845, + "kept_token_count": 571.9296875, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9909052371978759, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.552659800334368e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9018829345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.65028991699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.3011393964290619, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.13748923763632775, + "grad_norm": 0.0002436202921671793, + "kept_token_count": 571.7484375, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908093988895417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.590096566767898e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6724090576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7515472412109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28750, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29951933324337005, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.1359517253935337, + "grad_norm": 0.00018772538169287145, + "kept_token_count": 573.10625, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903737425804138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7602743032039146e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5177703857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1397735595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28760, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2992916852235794, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.12713682353496553, + "grad_norm": 0.0002722889475990087, + "kept_token_count": 573.5890625, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905709564685822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.683237355289748e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6938751220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6921417236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28770, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.2988982081413269, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.12832097932696343, + "grad_norm": 0.00019235306535847485, + "kept_token_count": 573.7484375, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9902965486049652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7904262717347593e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9321563720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.05238952636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28780, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2997035920619965, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.13168484717607498, + "grad_norm": 0.00025875793653540313, + "kept_token_count": 573.01875, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0, + "mgd_cos_sim": 0.99069904088974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6332026866148226e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6196014404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.13706970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28790, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30065402686595916, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.1324028566479683, + "grad_norm": 0.0001925894321175292, + "kept_token_count": 572.1546875, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990219521522522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.82051552151097e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1653656005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8822357177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29937092661857606, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.13216143995523452, + "grad_norm": 0.00022464121866505593, + "kept_token_count": 573.2421875, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908101499080658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.589805019146297e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.67486572265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8016784667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28810, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.3005509257316589, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.12423077002167701, + "grad_norm": 0.00025609874865040183, + "kept_token_count": 572.415625, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9908038973808289, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.592242665035883e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.48647155761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.8742706298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.30099949836730955, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.13114403411746026, + "grad_norm": 0.0002608439535833895, + "kept_token_count": 571.9890625, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905064761638641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.708426229422912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.263037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8053924560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30152629911899564, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.12544415444135665, + "grad_norm": 0.00022056352463550866, + "kept_token_count": 571.778125, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9904556214809418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.728288793354295e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6238555908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8174621582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28840, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.30037174820899964, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.1272055685520172, + "grad_norm": 0.00017606737674213946, + "kept_token_count": 572.353125, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906973659992218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.633861051639542e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5857849121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4125030517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28850, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30044410824775697, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.12024424374103546, + "grad_norm": 0.0002196138957515359, + "kept_token_count": 572.4515625, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9907699584960937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.605506535677705e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5978637695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.8233215332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28860, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3012821078300476, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.13238076716661454, + "grad_norm": 0.00022604908735956997, + "kept_token_count": 572.0375, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9901790022850037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8363473140634594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6701141357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7001617431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28870, + "valid_token_count": 818.6953125 + }, + { + "actual_token_mask_ratio": 0.3011169880628586, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.13511064872145653, + "grad_norm": 0.00023684004554525018, + "kept_token_count": 572.05625, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903157472610473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7829323264304546e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.25502319335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.83055725097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28880, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.2993656754493713, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.12930467128753662, + "grad_norm": 0.00019767948833759874, + "kept_token_count": 573.2078125, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905537605285645, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6899556653224864e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6801696777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0679534912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28890, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.3014595150947571, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.13422823697328568, + "grad_norm": 0.0003287200233899057, + "kept_token_count": 571.5390625, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9905382633209229, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6960088982596063e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1635406494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2053588867187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28900, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3007472395896912, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.12719108387827874, + "grad_norm": 0.0002782044466584921, + "kept_token_count": 572.265625, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990906172990799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5522882899385874e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9475799560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.17357482910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2993441253900528, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.12820345535874367, + "grad_norm": 0.00025220491806976497, + "kept_token_count": 573.371875, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9910403847694397, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.499869490042329e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0081420898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6792938232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28920, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.30087096989154816, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.13199664279818535, + "grad_norm": 0.0002197930298279971, + "kept_token_count": 572.034375, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9906866133213044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6380610436026475e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.34833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.53643493652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28930, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3004576534032822, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.12476506233215331, + "grad_norm": 0.00018717935017775744, + "kept_token_count": 572.440625, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0, + "mgd_cos_sim": 0.990760737657547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.609105078794528e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3272308349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6793487548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28940, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.3002090185880661, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.13256212025880815, + "grad_norm": 0.0002143828896805644, + "kept_token_count": 572.6359375, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0, + "mgd_cos_sim": 0.9903591394424438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.765980181924533e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.68902282714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2332824707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28950, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.30112705528736117, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.13478597328066827, + "grad_norm": 0.00021430951892398298, + "kept_token_count": 571.6546875, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9905787169933319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.680201516544912e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6942840576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.08229675292966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28960, + "valid_token_count": 817.971875 + }, + { + "actual_token_mask_ratio": 0.29911677837371825, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.12755366265773774, + "grad_norm": 0.00021662494691554457, + "kept_token_count": 573.5703125, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9903818726539612, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7570970380329524e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.30388793945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.91486206054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28970, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3008053183555603, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.1364045947790146, + "grad_norm": 0.00035043666139245033, + "kept_token_count": 571.975, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9905755281448364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.681445232359692e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.54197692871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.29501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28980, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.29953002333641054, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.11955187693238259, + "grad_norm": 0.0001551639725221321, + "kept_token_count": 572.909375, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9899811685085297, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.913625660061371e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.38720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.09429626464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28990, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.2981577664613724, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.12778663486242295, + "grad_norm": 0.0001880454656202346, + "kept_token_count": 574.103125, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9910655617713928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.490038579911925e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.41867980957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7253784179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29000, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.30033430457115173, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.1278632290661335, + "grad_norm": 0.0002592317759990692, + "kept_token_count": 572.3046875, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0, + "mgd_cos_sim": 0.990406745672226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.74738265236374e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0223693847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.14642028808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29010, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2995032429695129, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.1331437662243843, + "grad_norm": 0.0005814091418869793, + "kept_token_count": 573.3875, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901837527751922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8344896165654066e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8980407714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2467987060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29020, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.3004571348428726, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.12325963377952576, + "grad_norm": 0.00020458767539821565, + "kept_token_count": 572.08125, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0, + "mgd_cos_sim": 0.990349406003952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7697938751080073e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.844580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.12608032226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29030, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.30114546716213225, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.12661951780319214, + "grad_norm": 0.0002791336737573147, + "kept_token_count": 571.7140625, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907861173152923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5991894037579186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8384155273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8138427734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29040, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.3004530489444733, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.13767791613936425, + "grad_norm": 0.00021190434927120805, + "kept_token_count": 572.5953125, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9903611302375793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.765202964132186e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8481475830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6199432373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29050, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.299525710940361, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.1301637589931488, + "grad_norm": 0.0002692476555239409, + "kept_token_count": 572.9671875, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9902730286121368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.799618516495684e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.11036376953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.19190673828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29060, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.29976379573345185, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.13634562641382217, + "grad_norm": 0.00027386759757064283, + "kept_token_count": 572.934375, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0, + "mgd_cos_sim": 0.99021817445755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.821044447249733e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.079345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8332946777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29070, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.30063741505146024, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.13136355951428413, + "grad_norm": 0.00022667148732580245, + "kept_token_count": 572.4375, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9900268673896789, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.895777153957169e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3722198486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.90874938964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29080, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.29957905113697053, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.12766007483005523, + "grad_norm": 0.00021045462926849723, + "kept_token_count": 573.0828125, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909169316291809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5480954647937327e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.36396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36456298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29090, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2994110554456711, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.13286686167120934, + "grad_norm": 0.00018103429465554655, + "kept_token_count": 573.0390625, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906811535358429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6401964098331516e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0982604980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.32811584472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29100, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.30059218406677246, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.13273844495415688, + "grad_norm": 0.0002147281338693574, + "kept_token_count": 572.515625, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906502366065979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6522760638035834e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3599426269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6575531005859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29110, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3002567619085312, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.13094130903482437, + "grad_norm": 0.00022893844288773835, + "kept_token_count": 572.4984375, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9899829745292663, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.912927277269773e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1920654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.588232421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29120, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3002746939659119, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.12136247158050537, + "grad_norm": 0.0002503755677025765, + "kept_token_count": 572.5375, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9899582743644715, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.922560354112648e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0199462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.1629913330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29130, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3003591686487198, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.1295353963971138, + "grad_norm": 0.00018735746562015265, + "kept_token_count": 572.4875, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907214939594269, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6244391230866314e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.25000915527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.79310607910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29140, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3005173623561859, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.12278692871332168, + "grad_norm": 0.00022540072677657008, + "kept_token_count": 572.4171875, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0, + "mgd_cos_sim": 0.990363484621048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7642871393472885e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.14013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.77028198242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29150, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2994302809238434, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.12362204045057297, + "grad_norm": 0.00019894582510460168, + "kept_token_count": 573.09375, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909329235553741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5418384140939454e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2477996826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0168487548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29160, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.30050902962684634, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.12214386835694313, + "grad_norm": 0.0002611747768241912, + "kept_token_count": 572.3421875, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906477153301239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.653258354461286e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.01596984863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.85076904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29170, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30076347291469574, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.13706002980470658, + "grad_norm": 0.00017747306264936924, + "kept_token_count": 572.1953125, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906230211257935, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.662905182864051e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9127716064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9781951904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29180, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.3002448707818985, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.12884612306952475, + "grad_norm": 0.00020738522289320827, + "kept_token_count": 572.675, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9900020003318787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.905486337316688e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8908172607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3378509521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29190, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29995912313461304, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.11775219291448594, + "grad_norm": 0.00017368255066685379, + "kept_token_count": 572.725, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909907460212708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.519263591442723e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9294372558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6530395507813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29200, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2997808426618576, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.12434471175074577, + "grad_norm": 0.00028366773040033877, + "kept_token_count": 573.103125, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0, + "mgd_cos_sim": 0.991207730770111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.434505379118491e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.61390991210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0232360839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29210, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.29905062913894653, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.12745836824178697, + "grad_norm": 0.0002314976300112903, + "kept_token_count": 573.590625, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9908186376094819, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.586481634556549e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.98305358886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.12090454101565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29220, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3000272184610367, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.13286497071385384, + "grad_norm": 0.000256047846050933, + "kept_token_count": 572.725, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0, + "mgd_cos_sim": 0.990522038936615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7023495315224865e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0522735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.92222290039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3014880269765854, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.13251537904143335, + "grad_norm": 0.0002617258869577199, + "kept_token_count": 572.028125, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906731903553009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.643303571152501e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.59552612304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7258239746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29240, + "valid_token_count": 818.9203125 + }, + { + "actual_token_mask_ratio": 0.2995980024337769, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.11964808329939843, + "grad_norm": 0.00019749626517295837, + "kept_token_count": 573.115625, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9908331155776977, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.580839911592193e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.48211059570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8088806152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29250, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.300596809387207, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.13247668743133545, + "grad_norm": 0.000314337172312662, + "kept_token_count": 572.3328125, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909992694854737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.51592652805266e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.68602600097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1429077148438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29260, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.30046742260456083, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.1269769087433815, + "grad_norm": 0.00021838060638401657, + "kept_token_count": 572.3625, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906786918640137, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6411556538951115e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.62254638671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3704528808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29270, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3008282959461212, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.12606247141957283, + "grad_norm": 0.0002971309586428106, + "kept_token_count": 572.1625, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9903219819068909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7804913517902605e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.757958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.00002136230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29280, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29876691699028013, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.13508933559060096, + "grad_norm": 0.0002301347121829167, + "kept_token_count": 573.5640625, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9900894343852997, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8713305548299105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.60866088867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.49420166015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29290, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.2997831881046295, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.12186161801218987, + "grad_norm": 0.00023527309531345963, + "kept_token_count": 572.85625, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9903544247150421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7678196167689745e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.68780517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.47413024902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29300, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.299466010928154, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.13498205989599227, + "grad_norm": 0.0003094962448813021, + "kept_token_count": 573.34375, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9897928416728974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.9871878107078375e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5434234619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8730529785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29310, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2996750295162201, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.13375124633312224, + "grad_norm": 0.00023227365454658866, + "kept_token_count": 573.0375, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906061708927154, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.669484904094134e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4715942382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.24033508300784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29320, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3000865250825882, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.11807586699724197, + "grad_norm": 0.00023525298456661403, + "kept_token_count": 572.6953125, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9902000963687897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8281142769847067e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.324169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.65940856933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29330, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2999401718378067, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.1263599045574665, + "grad_norm": 0.00016766690532676876, + "kept_token_count": 573.003125, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909982144832611, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.516340038913768e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.25432739257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1955505371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2991983026266098, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.13879528641700745, + "grad_norm": 0.00027967215282842517, + "kept_token_count": 573.4875, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9908975243568421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.555672556103673e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.17249450683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8886688232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3005922377109528, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.13172420635819435, + "grad_norm": 0.00020887576101813465, + "kept_token_count": 572.2265625, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901708364486694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.839532582787797e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.52228088378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5160888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29360, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3001681089401245, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.12844740673899652, + "grad_norm": 0.00022566496045328677, + "kept_token_count": 572.9140625, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0, + "mgd_cos_sim": 0.990190988779068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.831664798781276e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2625427246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3422576904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29370, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.299747970700264, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.12536565437912942, + "grad_norm": 0.00020664476323872805, + "kept_token_count": 573.1015625, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9905483484268188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.692070840770612e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.70498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.509912109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29380, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2999190390110016, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.12769917845726014, + "grad_norm": 0.0002292936114827171, + "kept_token_count": 572.725, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9905695140361785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.683803370222449e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.20384216308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7363616943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29390, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30046661794185636, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.12108035907149314, + "grad_norm": 0.00021055107936263084, + "kept_token_count": 572.3609375, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907261013984681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.622633848863188e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4501678466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5986053466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29400, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29972840249538424, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.1217944636940956, + "grad_norm": 0.00015634445298928767, + "kept_token_count": 573.0453125, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9911597967147827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.4532200152170846e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.63856811523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.5445831298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29410, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30106774568557737, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.13303395360708237, + "grad_norm": 0.0001812406990211457, + "kept_token_count": 571.8046875, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907105088233947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.628728481999133e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6245483398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4945007324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29420, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.3005947768688202, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.1403844691812992, + "grad_norm": 0.00019345045438967645, + "kept_token_count": 572.553125, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9900757551193238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.876677110383753e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9297668457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6683624267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29430, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.3001375585794449, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.11950915157794953, + "grad_norm": 0.00020139482512604445, + "kept_token_count": 572.515625, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907721102237701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.604662779252976e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0653350830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9148223876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29440, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30022660195827483, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.12797450199723243, + "grad_norm": 0.0001962795649887994, + "kept_token_count": 572.6171875, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907221376895905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.624186647357419e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2208953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8215637207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29450, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29983108937740327, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.1386633411049843, + "grad_norm": 0.00026687554782256484, + "kept_token_count": 573.0015625, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901534438133239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8463295277324504e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9129638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6815124511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29460, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2997604101896286, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.13452932462096215, + "grad_norm": 0.0003051422827411443, + "kept_token_count": 573.0125, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907632410526276, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608137012633961e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.26884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.07237854003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29470, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29950241148471834, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.12828799411654473, + "grad_norm": 0.00020481247338466346, + "kept_token_count": 573.1109375, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901685595512391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.84042603400303e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2917175292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.973095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29480, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.3009018003940582, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.12878162115812303, + "grad_norm": 0.00019079187768511474, + "kept_token_count": 572.0921875, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9904370307922363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.735551508725621e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1943756103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3500671386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29490, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.29989061653614046, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.12672675549983978, + "grad_norm": 0.0002303497021785006, + "kept_token_count": 572.734375, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0, + "mgd_cos_sim": 0.989702707529068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 4.022402754344512e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.69288940429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.97687072753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29500, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.30064386427402495, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.12750406563282013, + "grad_norm": 0.00020719875465147197, + "kept_token_count": 572.41875, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907634973526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608027127484092e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.60857849121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.43024291992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29510, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.29896336793899536, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.13100989609956742, + "grad_norm": 0.0002772344450931996, + "kept_token_count": 573.3703125, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901402771472931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.851472538372036e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.49924926757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7959808349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29520, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.2996380031108856, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.12994176149368286, + "grad_norm": 0.00021633246797136962, + "kept_token_count": 572.9609375, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9907014667987823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.632254920375999e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.50185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6781890869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29530, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2993350297212601, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.1335747145116329, + "grad_norm": 0.00027659759507514536, + "kept_token_count": 573.2765625, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9903824090957641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.756891092052683e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.55360412597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01255798339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29540, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3002115964889526, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.12249992266297341, + "grad_norm": 0.0002322782383998856, + "kept_token_count": 572.69375, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9899057924747467, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.943073497794103e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8523834228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3866790771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29550, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29956406354904175, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.13085950687527656, + "grad_norm": 0.0002639738377183676, + "kept_token_count": 573.1765625, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9904962778091431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7124050868442283e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.79442138671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.09642333984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29560, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2996412605047226, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.12923907339572907, + "grad_norm": 0.0002574519021436572, + "kept_token_count": 573.03125, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9908011853694916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.593303354136879e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8014709472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.88161315917966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29570, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3006107062101364, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.1354766868054867, + "grad_norm": 0.00025385821936652064, + "kept_token_count": 572.3875, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9905142247676849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7053992855362594e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5899688720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.44536743164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29580, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.30006812512874603, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.1291450448334217, + "grad_norm": 0.00025883252965286374, + "kept_token_count": 572.8078125, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9904488503932953, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7309395702322944e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.31297607421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4328155517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29590, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30065059661865234, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.1344200387597084, + "grad_norm": 0.00022820105368737131, + "kept_token_count": 572.125, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9901081383228302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.864030695694964e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9674041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2766937255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29600, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2995491087436676, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.1272275298833847, + "grad_norm": 0.0003875689289998263, + "kept_token_count": 573.2515625, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9902115166187286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.823652004939504e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.88926086425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.66840209960935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29610, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.3006875902414322, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.13308432474732398, + "grad_norm": 0.000190502280020155, + "kept_token_count": 572.2296875, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9900476217269898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.88766398828011e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5780029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.619580078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29620, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30075899958610536, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.12869151532649994, + "grad_norm": 0.00019668351160362363, + "kept_token_count": 572.121875, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9906626999378204, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.647400772024412e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2074310302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20184020996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29630, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2996909350156784, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.12944766655564308, + "grad_norm": 0.00031361094443127513, + "kept_token_count": 572.9515625, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0, + "mgd_cos_sim": 0.99086292386055, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.569192158465739e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4920166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6311737060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29640, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.3002570331096649, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.12612590417265893, + "grad_norm": 0.00020776237943209708, + "kept_token_count": 572.684375, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9904419422149658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7336324385250916e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7827514648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.53464965820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29650, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.29974748492240905, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.1273024469614029, + "grad_norm": 0.00023579820117447525, + "kept_token_count": 573.1015625, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0, + "mgd_cos_sim": 0.9909698665142059, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.527421358739957e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.66884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.57916870117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29660, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.29881702065467836, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.12199114337563514, + "grad_norm": 0.00024115643464028835, + "kept_token_count": 573.85, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9907137334346772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6274693411542105e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.585888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1617889404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29670, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3007679104804993, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.12495990321040154, + "grad_norm": 0.0001827503729145974, + "kept_token_count": 572.1671875, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9903061211109161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7866907950956376e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6381011962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.83534545898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29680, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3007578909397125, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.12837121561169623, + "grad_norm": 0.000204074356588535, + "kept_token_count": 572.1609375, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9907632172107697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.608137449191417e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.495263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.37099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30023435056209563, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.12492757886648179, + "grad_norm": 0.00023136204981710762, + "kept_token_count": 572.68125, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0, + "mgd_cos_sim": 0.990436440706253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.7357772089308125e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.85686950683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1148345947266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29700, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29989232420921325, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.1289367601275444, + "grad_norm": 0.00022829772206023335, + "kept_token_count": 572.8484375, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9903534412384033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.768201931961812e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9513885498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1768493652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3004705309867859, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.12210648506879807, + "grad_norm": 0.0001965317060239613, + "kept_token_count": 572.425, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9912316858768463, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.425138602324296e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6343048095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9423461914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29720, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.30102503299713135, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.12663280442357064, + "grad_norm": 0.00022136713960208, + "kept_token_count": 571.975, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0, + "mgd_cos_sim": 0.990644347667694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.654571155493613e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4607666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.54738464355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.29914834499359133, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.12205351665616035, + "grad_norm": 0.00029631759389303625, + "kept_token_count": 573.6703125, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9908133387565613, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.588565377867781e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7737548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9098815917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29740, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.2998922854661942, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.12063553258776664, + "grad_norm": 0.00021997932344675064, + "kept_token_count": 573.0640625, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9901676177978516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.840789104287978e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2351318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.84918518066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29750, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.2999304383993149, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.1312817871570587, + "grad_norm": 0.00016522937221452594, + "kept_token_count": 572.9203125, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9907963991165161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.595171328925062e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2974487304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4618377685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29760, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2999958127737045, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.13024647161364555, + "grad_norm": 0.00022105470998212695, + "kept_token_count": 572.778125, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9904304563999176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.738120176421944e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.08233642578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1991424560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.30076673328876496, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.12700564935803413, + "grad_norm": 0.0002068772737402469, + "kept_token_count": 572.184375, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9902376294136047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.813442453974858e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2284729003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.98907775878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29780, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2997681379318237, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.1237889938056469, + "grad_norm": 0.00024816617951728404, + "kept_token_count": 572.9078125, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9909230172634125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5457122066873126e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.05755004882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.6459533691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29790, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.3007602095603943, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.13288340643048285, + "grad_norm": 0.00016173915355466306, + "kept_token_count": 571.9671875, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9907078325748444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.629767961683683e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.78289794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0460540771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29800, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.3008357107639313, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.1187626525759697, + "grad_norm": 0.00019299788982607424, + "kept_token_count": 572.084375, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0, + "mgd_cos_sim": 0.990850955247879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.573870526452083e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1819641113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9902374267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29810, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2991673588752747, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.12125391513109207, + "grad_norm": 0.00022828459623269737, + "kept_token_count": 573.3421875, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9904078185558319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.746969778148923e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4995056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.36883850097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29820, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.3010241210460663, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.13129108548164367, + "grad_norm": 0.00023709530069027096, + "kept_token_count": 572.0125, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9908797800540924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.562605706974864e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.87626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.969873046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29830, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.29824419021606446, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.13195696100592613, + "grad_norm": 0.00019541129586286843, + "kept_token_count": 574.1578125, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9905329465866088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.69808552932227e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.27971801757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.59498291015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29840, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30011793673038484, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.13681414276361464, + "grad_norm": 0.00016846382641233504, + "kept_token_count": 572.609375, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9904642999172211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.72490339941578e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.88124694824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.02699890136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29850, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.30031400322914126, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.1363554522395134, + "grad_norm": 0.00019407244690228254, + "kept_token_count": 572.7140625, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9913606822490693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.374753177922685e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.56654663085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.5957763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29860, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.3012273073196411, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.12710360288619996, + "grad_norm": 0.000213874620385468, + "kept_token_count": 571.6765625, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9909349620342255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5410495183896276e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.10057373046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.26352844238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29870, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2986297786235809, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.12256635501980781, + "grad_norm": 0.0002273078862344846, + "kept_token_count": 573.8265625, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9908180832862854, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.5867037149728274e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.57268676757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.84569396972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29880, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2985735833644867, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.13200969249010086, + "grad_norm": 0.00020121106354054064, + "kept_token_count": 573.9109375, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0, + "mgd_cos_sim": 0.9910822927951812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.483494401734788e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.04184875488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.42064208984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29890, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3005139023065567, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.1265594057738781, + "grad_norm": 0.00023776458692736924, + "kept_token_count": 572.53125, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9902713537216187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.8002723158570006e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4851837158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29900, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.2998731195926666, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.1300799310207367, + "grad_norm": 0.00017145440506283194, + "kept_token_count": 573.0421875, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9906623542308808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.647538087534485e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2151336669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.29319763183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29910, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.2988147884607315, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.12547900676727294, + "grad_norm": 0.00027558140573091805, + "kept_token_count": 573.6515625, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9905897200107574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6759147042175756e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.11423950195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4685791015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29920, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30034813582897185, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.1342018686234951, + "grad_norm": 0.0002688105159904808, + "kept_token_count": 572.4078125, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9897847652435303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.99033997382503e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3507476806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1698425292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29930, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.29917544424533843, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.13313272297382356, + "grad_norm": 0.000287779257632792, + "kept_token_count": 573.4828125, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9906591653823853, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.648779857030604e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0418762207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1836669921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29940, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3017361551523209, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.13007839024066925, + "grad_norm": 0.0002261391346110031, + "kept_token_count": 571.378125, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9901519596576691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.846904473903123e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7433288574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0750274658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2997448056936264, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.13431318253278732, + "grad_norm": 0.0002369807771174237, + "kept_token_count": 572.984375, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0, + "mgd_cos_sim": 0.9905858933925629, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.677405620692298e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0935302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.4425933837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29960, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30060445666313174, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.13656316921114922, + "grad_norm": 0.00023541069822385907, + "kept_token_count": 572.1421875, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0, + "mgd_cos_sim": 0.9907787024974823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.60208747224533e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6768493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9291961669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29970, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.2996755510568619, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.1283566474914551, + "grad_norm": 0.0002648363006301224, + "kept_token_count": 572.8578125, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0, + "mgd_cos_sim": 0.9906814634799957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.640077593445312e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.51268005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6571533203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29980, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.29970529675483704, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.12324222698807716, + "grad_norm": 0.00021205142547842115, + "kept_token_count": 573.15625, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0, + "mgd_cos_sim": 0.9910810589790344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.483980490273098e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.66282348632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2133819580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29990, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30140580236911774, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.12389749586582184, + "grad_norm": 0.00020926726574543864, + "kept_token_count": 571.8390625, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0, + "mgd_cos_sim": 0.9905681610107422, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 3.6843320049229075e-05, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9220764160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7536865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30000, + "valid_token_count": 818.5625 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 9.218714250503885e-05, + "train_runtime": 56629.406, + "train_samples_per_second": 67.809, + "train_steps_per_second": 0.53 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..edc8c7c25949953cb796ba2aa6e2d1dafce8fc12 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase2/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36b22b1e779cb8364f92308094b4e4181a65c15557d955edb39862363a6d3911 +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/config.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f7932ab86992cc43c7506a0dc1ec8756a609a19a --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": true, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.5, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": null, + "mgd_loss_weight_start": 0.05, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.3, + "mgd_use_cosine_loss": false, + "mgd_use_mse_loss": true, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b0dc501198f581dc5126cae26dcd2d1233823eeb --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b01d8089bcfe5112dcbcb61d3a53975724cde989df4cdbf3fae480dd24f2e55 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..524c1787376355a3d299729688c23f5f51ca4c52 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:accb04d1a46feada7cfcbbcca4135cbe828f5e84672ad66c2832a6bf2a11fe37 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/trainer_state.json b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..62e15e0a13bf746a8bad1e924fceabfc1e6facaa --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.29861474335193633, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.14177069440484047, + "grad_norm": 0.6304369568824768, + "kept_token_count": 574.103125, + "learning_rate": 1.8e-07, + "loss": 0.1418, + "mgd_cos_sim": 0.9896599054336548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 4.039121231471654e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1236114501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6088409423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.300568500161171, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.13743472695350648, + "grad_norm": 0.785057544708252, + "kept_token_count": 572.3640625, + "learning_rate": 3.8e-07, + "loss": 0.1375, + "mgd_cos_sim": 0.9907497346401215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.6134002584731206e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.09418334960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8645324707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.29994507133960724, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.12628115341067314, + "grad_norm": 0.7366859316825867, + "kept_token_count": 572.903125, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1263, + "mgd_cos_sim": 0.9904629170894623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.7254435301292685e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.66595764160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7368560791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2992320328950882, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.10593516901135444, + "grad_norm": 0.5440253615379333, + "kept_token_count": 573.590625, + "learning_rate": 7.799999999999999e-07, + "loss": 0.106, + "mgd_cos_sim": 0.9906616687774659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.647802623163443e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.6528594970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5114379882813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 40, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.30005078911781313, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.09310055524110794, + "grad_norm": 0.5714760422706604, + "kept_token_count": 572.7359375, + "learning_rate": 9.8e-07, + "loss": 0.0931, + "mgd_cos_sim": 0.9909688472747803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.5278124050819316e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.8240264892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2966827392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 50, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30025090277194977, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.07936431840062141, + "grad_norm": 0.4794594347476959, + "kept_token_count": 572.4890625, + "learning_rate": 1.18e-06, + "loss": 0.0794, + "mgd_cos_sim": 0.9897827208042145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.991139274148736e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.1928283691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5880584716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 60, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30025394558906554, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.06545762941241265, + "grad_norm": 0.538167417049408, + "kept_token_count": 572.678125, + "learning_rate": 1.38e-06, + "loss": 0.0655, + "mgd_cos_sim": 0.990112566947937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.862302437482867e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.1177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6990570068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 70, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.30041784346103667, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.04744470305740833, + "grad_norm": 0.3662911653518677, + "kept_token_count": 572.4640625, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0475, + "mgd_cos_sim": 0.9902535974979401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 3.80720983230276e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.76519165039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.7150482177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 80, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29882457852363586, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.040651338174939156, + "grad_norm": 0.3197880685329437, + "kept_token_count": 573.640625, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0407, + "mgd_cos_sim": 0.9884834349155426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 4.498675007198471e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.4956024169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.26446838378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 90, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3003698617219925, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.03296433407813311, + "grad_norm": 0.20501238107681274, + "kept_token_count": 572.559375, + "learning_rate": 1.98e-06, + "loss": 0.033, + "mgd_cos_sim": 0.9885432302951813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 4.475319219636731e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.50427703857423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.25360412597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 100, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.300407937169075, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.03226143643260002, + "grad_norm": 0.19311928749084473, + "kept_token_count": 572.3265625, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0323, + "mgd_cos_sim": 0.9881967663764953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 4.610655923897866e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.05306396484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3141754150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 110, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.3005237102508545, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.025524352863430976, + "grad_norm": 0.1711236834526062, + "kept_token_count": 572.29375, + "learning_rate": 2.38e-06, + "loss": 0.0255, + "mgd_cos_sim": 0.9874474406242371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 4.9033631148631686e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.50356750488282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.33469848632814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 120, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29920447468757627, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.02324008271098137, + "grad_norm": 0.16659687459468842, + "kept_token_count": 573.18125, + "learning_rate": 2.58e-06, + "loss": 0.0233, + "mgd_cos_sim": 0.9869586944580078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 5.094278203614522e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 171.1873977661133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.1365692138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 130, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.2993192344903946, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.02449031174182892, + "grad_norm": 0.13082319498062134, + "kept_token_count": 573.4015625, + "learning_rate": 2.78e-06, + "loss": 0.0245, + "mgd_cos_sim": 0.9868053555488586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 5.154180435056333e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.51968536376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.33351745605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30067590773105624, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.02466187849640846, + "grad_norm": 0.161330908536911, + "kept_token_count": 572.18125, + "learning_rate": 2.98e-06, + "loss": 0.0247, + "mgd_cos_sim": 0.9859562158584595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 5.4858700968907216e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.176806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.651220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 150, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3000741422176361, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.022137156780809165, + "grad_norm": 0.14936678111553192, + "kept_token_count": 572.7796875, + "learning_rate": 3.18e-06, + "loss": 0.0222, + "mgd_cos_sim": 0.984583956003189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.021914014127105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.96834716796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.45213623046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 160, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2999145835638046, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.01977353570982814, + "grad_norm": 0.12585368752479553, + "kept_token_count": 572.6546875, + "learning_rate": 3.38e-06, + "loss": 0.0198, + "mgd_cos_sim": 0.9842170834541321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.165227678138763e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.00433197021485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.11622924804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.3002263009548187, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.017560439463704825, + "grad_norm": 0.14171098172664642, + "kept_token_count": 572.7140625, + "learning_rate": 3.58e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.984695291519165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 5.9784253971884026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.54931640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8096649169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 180, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3003869980573654, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.018752070609480143, + "grad_norm": 0.13648274540901184, + "kept_token_count": 572.6109375, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9837572395801544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.344846806314309e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 105.4951400756836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.06630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 190, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.2999288201332092, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.018863328266888857, + "grad_norm": 0.13390181958675385, + "kept_token_count": 572.796875, + "learning_rate": 3.98e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9830672979354859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614358062506653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 101.19379577636718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.9151336669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 200, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29890665113925935, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.018878384865820406, + "grad_norm": 0.11192018538713455, + "kept_token_count": 573.9203125, + "learning_rate": 4.18e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9812065601348877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.341209711739794e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.05995864868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.06500244140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 210, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.30033319890499116, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.018056638445705174, + "grad_norm": 0.11191882938146591, + "kept_token_count": 572.215625, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0181, + "mgd_cos_sim": 0.982239693403244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.937636717339047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.6418342590332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.256005859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 220, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.29963326156139375, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.020296731591224672, + "grad_norm": 0.13157989084720612, + "kept_token_count": 572.840625, + "learning_rate": 4.58e-06, + "loss": 0.0203, + "mgd_cos_sim": 0.9808040499687195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.498441264033317e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72452774047852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5458679199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 230, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.2992518484592438, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.022011865116655826, + "grad_norm": 0.13000856339931488, + "kept_token_count": 573.1890625, + "learning_rate": 4.78e-06, + "loss": 0.022, + "mgd_cos_sim": 0.980921620130539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.452512290910818e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25609588623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2445434570312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 240, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.2995978981256485, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.017801968939602376, + "grad_norm": 0.12953320145606995, + "kept_token_count": 573.3171875, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.979910159111023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.847609158488922e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.3116439819336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0098876953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 250, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.30087141692638397, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.017109345085918902, + "grad_norm": 0.13886740803718567, + "kept_token_count": 572.0921875, + "learning_rate": 5.18e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9808664739131927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.47405327274464e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.52200469970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0119598388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 260, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3005953520536423, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.018411026895046236, + "grad_norm": 0.091580331325531, + "kept_token_count": 572.1921875, + "learning_rate": 5.38e-06, + "loss": 0.0185, + "mgd_cos_sim": 0.9787208497524261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.312189966090955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.1246841430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3732452392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 270, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3004117637872696, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.016209184285253288, + "grad_norm": 0.10805875062942505, + "kept_token_count": 572.340625, + "learning_rate": 5.58e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9793108463287353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.081720661721193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.99839096069336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.18971252441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 280, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.3009315937757492, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.015993032325059175, + "grad_norm": 0.10148482769727707, + "kept_token_count": 571.9234375, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9800072014331818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.809704620740377e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.85706024169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.37989501953126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 290, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29988266825675963, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.01875318670645356, + "grad_norm": 0.10372697561979294, + "kept_token_count": 572.6828125, + "learning_rate": 5.98e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9779353857040405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.619008585810661e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.05033874511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.82794189453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 300, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.2990699976682663, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.01985787320882082, + "grad_norm": 0.09332051128149033, + "kept_token_count": 573.5984375, + "learning_rate": 6.18e-06, + "loss": 0.0199, + "mgd_cos_sim": 0.9783569276332855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.454342532786541e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.685791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.08152160644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2991314947605133, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.0138938894495368, + "grad_norm": 0.11711727827787399, + "kept_token_count": 573.4625, + "learning_rate": 6.38e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9781495094299316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.535368833690881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.37147521972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.80592346191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 320, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.30076790452003477, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.01813919097185135, + "grad_norm": 0.10911791771650314, + "kept_token_count": 572.05625, + "learning_rate": 6.58e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9777928769588471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.674674900248647e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.93936157226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.28992614746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 330, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2996584057807922, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.02089148610830307, + "grad_norm": 0.10991784930229187, + "kept_token_count": 573.1, + "learning_rate": 6.78e-06, + "loss": 0.0209, + "mgd_cos_sim": 0.9764366090297699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.204470334225334e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.65210456848145, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.531201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 340, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2997130542993546, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.01714860089123249, + "grad_norm": 0.1258341372013092, + "kept_token_count": 572.821875, + "learning_rate": 6.98e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9778466284275055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.653681725263595e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.49641036987305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.949072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 350, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.29932459592819216, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.0194732409901917, + "grad_norm": 0.0856800228357315, + "kept_token_count": 573.521875, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0195, + "mgd_cos_sim": 0.9768641769886017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.037449417519383e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.78607597351074, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.60358276367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 360, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.30039571821689603, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.016617572354152798, + "grad_norm": 0.089473195374012, + "kept_token_count": 572.58125, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9764897525310516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.183707807096653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.27258720397949, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3290679931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 370, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.3009997606277466, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.016165194846689702, + "grad_norm": 0.10490547120571136, + "kept_token_count": 571.9234375, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.975509113073349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.566765947965905e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.49661560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8820556640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 380, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3000940829515457, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.015313808340579271, + "grad_norm": 0.08932042866945267, + "kept_token_count": 572.590625, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9765994012355804, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.140881156781689e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.541999435424806, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.5934753417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 390, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.3006135135889053, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.016488548275083302, + "grad_norm": 0.09324696660041809, + "kept_token_count": 572.2375, + "learning_rate": 7.98e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9766215920448303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.132212871918455e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.805472946166994, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5403137207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 400, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30034646987915037, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.017593976203352214, + "grad_norm": 0.09754008054733276, + "kept_token_count": 572.3453125, + "learning_rate": 8.18e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9764261901378631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.208540213876404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.09891357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.9258544921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 410, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.29918642342090607, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.016376363579183817, + "grad_norm": 0.07628296315670013, + "kept_token_count": 573.4640625, + "learning_rate": 8.38e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9758258163928986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.4430563331116e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.31959991455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1865478515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 420, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29940618872642516, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.019465667940676213, + "grad_norm": 0.13221602141857147, + "kept_token_count": 573.1984375, + "learning_rate": 8.58e-06, + "loss": 0.0195, + "mgd_cos_sim": 0.9736301124095916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010300753201590851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.14254302978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7475860595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3006904125213623, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.01619293764233589, + "grad_norm": 0.11691811680793762, + "kept_token_count": 572.43125, + "learning_rate": 8.78e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9754431426525116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.592540300218388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.06342658996582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5489532470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 440, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.300125727057457, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.015663806162774564, + "grad_norm": 0.09612502157688141, + "kept_token_count": 572.6046875, + "learning_rate": 8.98e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9760525166988373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.354503999929875e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.94430084228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7301361083984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.300143826007843, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.01538244280964136, + "grad_norm": 0.1069297268986702, + "kept_token_count": 572.6625, + "learning_rate": 9.18e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9759531438350677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.393322688993067e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.83189697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.04462280273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 460, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3000409543514252, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.014898433536291122, + "grad_norm": 0.09041168540716171, + "kept_token_count": 572.8078125, + "learning_rate": 9.38e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9738198220729828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0001022665594064165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.41831321716309, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.79323425292966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 470, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2990784883499146, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.01655669594183564, + "grad_norm": 0.1217784732580185, + "kept_token_count": 573.65625, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9745527803897858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.940338641172275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.95564918518066, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.99134826660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 480, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.30011009275913236, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.01610007956624031, + "grad_norm": 0.13140426576137543, + "kept_token_count": 572.7828125, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9741188824176789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010109833820024506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.912635803222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.39647521972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 490, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29960916340351107, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.015693806018680333, + "grad_norm": 0.09759970754384995, + "kept_token_count": 572.9015625, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9755480825901032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.551548428134992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.60665321350098, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.51275329589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 500, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.3008320689201355, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.013541182689368724, + "grad_norm": 0.10469485819339752, + "kept_token_count": 572.0625, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9749751091003418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.77536546997726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.38908462524414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.40911560058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 510, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.30012530982494356, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.016928760800510645, + "grad_norm": 0.11418323218822479, + "kept_token_count": 572.759375, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.017, + "mgd_cos_sim": 0.973829197883606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010222993150819094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.55037612915039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4741516113281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 520, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2998683929443359, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.016500164568424226, + "grad_norm": 0.12838687002658844, + "kept_token_count": 572.846875, + "learning_rate": 1.058e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.974366021156311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0001001329357677605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.17152824401855, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3198577880859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 530, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30005150437355044, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.01808418333530426, + "grad_norm": 0.11730771511793137, + "kept_token_count": 572.684375, + "learning_rate": 1.078e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9735597252845765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010328245116397738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.82263259887695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.284521484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 540, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29973108172416685, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.013957243878394365, + "grad_norm": 0.10311868786811829, + "kept_token_count": 573.1984375, + "learning_rate": 1.098e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9730113446712494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010542461095610633, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.62228775024414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.40354919433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 550, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.29996989369392396, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.017979083210229875, + "grad_norm": 0.11572984606027603, + "kept_token_count": 572.671875, + "learning_rate": 1.118e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9745990991592407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.922245953930542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.70473670959473, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6068481445312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 560, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29887217581272124, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.017269617505371572, + "grad_norm": 0.10523637384176254, + "kept_token_count": 573.6609375, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.975819218158722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.44563937082421e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.46583023071289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.17588195800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 570, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.29946407675743103, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.01677524168044329, + "grad_norm": 0.12330259382724762, + "kept_token_count": 573.0640625, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9748232126235962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.834705197135918e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.21952171325684, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.600341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 580, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.2999726623296738, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.017119337804615498, + "grad_norm": 0.09189106523990631, + "kept_token_count": 572.8, + "learning_rate": 1.178e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.973579329252243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010320587971364148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.59206428527832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4979644775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 590, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29969131052494047, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.016811166144907474, + "grad_norm": 0.10206809639930725, + "kept_token_count": 573.1, + "learning_rate": 1.198e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9752906501293183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.652110747992992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 49.91346969604492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9354736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 600, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30001686215400697, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.01546208206564188, + "grad_norm": 0.12289480865001678, + "kept_token_count": 572.6765625, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9751817464828492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.694647233118303e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.6178840637207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.6006286621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 610, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30020990371704104, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.015686834510415793, + "grad_norm": 0.10005412995815277, + "kept_token_count": 572.4609375, + "learning_rate": 1.238e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9747754395008087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.853365481831134e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.798469161987306, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0475769042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 620, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.300502148270607, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.01778702074661851, + "grad_norm": 0.08330614119768143, + "kept_token_count": 572.453125, + "learning_rate": 1.258e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9750290393829346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.754302009241655e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.52594184875488, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.7279479980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 630, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.3001162141561508, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.01722151767462492, + "grad_norm": 0.10143925249576569, + "kept_token_count": 572.6296875, + "learning_rate": 1.278e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9757688462734222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.465307884966023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.700252532958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.82276306152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 640, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.30013062357902526, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.015600314363837242, + "grad_norm": 0.09705853462219238, + "kept_token_count": 572.575, + "learning_rate": 1.298e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9761093437671662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.33230621740222e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.606082916259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.03101501464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 650, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.3007644057273865, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.01691175177693367, + "grad_norm": 0.11178331822156906, + "kept_token_count": 572.115625, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9757005870342255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.491977471043356e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.885684204101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9715209960938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 660, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2995691657066345, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.015615269541740417, + "grad_norm": 0.09902986884117126, + "kept_token_count": 573.228125, + "learning_rate": 1.338e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9763279616832733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.246904228348285e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.75507354736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.02466735839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 670, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2992744654417038, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.015194008871912957, + "grad_norm": 0.11417253315448761, + "kept_token_count": 573.3359375, + "learning_rate": 1.358e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9760403573513031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.359255127492361e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.16379165649414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9166748046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 680, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29973390102386477, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.01375067294575274, + "grad_norm": 0.08501788973808289, + "kept_token_count": 573.0671875, + "learning_rate": 1.378e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9761382818222046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.32100614591036e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.72579498291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6299560546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 690, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30091370046138766, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.015389035269618035, + "grad_norm": 0.12585775554180145, + "kept_token_count": 572.215625, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9758432686328888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.436242617084644e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.8708251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3120971679688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 700, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2991113871335983, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.01577328685671091, + "grad_norm": 0.10644292086362839, + "kept_token_count": 573.75, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9756433129310608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.5143482030835e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.246002960205075, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.587451171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 710, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.2999574333429337, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.015158306621015073, + "grad_norm": 0.08158517628908157, + "kept_token_count": 572.6453125, + "learning_rate": 1.438e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9756839275360107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.498479194007814e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.62632026672363, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.38450317382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 720, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.3000538289546967, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.015217255800962448, + "grad_norm": 0.0992538258433342, + "kept_token_count": 572.8046875, + "learning_rate": 1.458e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.975578510761261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.53965623921249e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.791238403320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.48280639648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 730, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30041956901550293, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.017550715152174233, + "grad_norm": 0.10385022312402725, + "kept_token_count": 572.4046875, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9748896479606628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.808748291106895e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.7572582244873, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.15802612304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 740, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2988370984792709, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.017811464983969925, + "grad_norm": 0.14324797689914703, + "kept_token_count": 573.8265625, + "learning_rate": 1.498e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9761492490768433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.316716168541461e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.05026435852051, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9805847167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 750, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2989182144403458, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.016667833365499974, + "grad_norm": 0.1476774513721466, + "kept_token_count": 573.5734375, + "learning_rate": 1.518e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9763843953609467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.224861933034844e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.617654418945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.55238952636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 760, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2994613766670227, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.019402474630624056, + "grad_norm": 0.13414616882801056, + "kept_token_count": 573.0765625, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.974442994594574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.983222553273663e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.103875732421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4068908691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 770, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.3002647817134857, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.016429266054183246, + "grad_norm": 0.10504431277513504, + "kept_token_count": 572.4390625, + "learning_rate": 1.558e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9756853938102722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.497907813056372e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.960253143310545, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6106811523438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 780, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2995858073234558, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.014357122872024775, + "grad_norm": 0.13077589869499207, + "kept_token_count": 572.80625, + "learning_rate": 1.578e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9764833629131318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.186200841213577e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.29398078918457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.7010528564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 790, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.2997112959623337, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.01692212522029877, + "grad_norm": 0.11311578750610352, + "kept_token_count": 572.803125, + "learning_rate": 1.598e-05, + "loss": 0.017, + "mgd_cos_sim": 0.975842010974884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.436733307666145e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.32690315246582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2421020507812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 800, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.29949976205825807, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.013229265343397856, + "grad_norm": 0.07774095982313156, + "kept_token_count": 573.128125, + "learning_rate": 1.618e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9767671227455139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.075368434423581e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.76585235595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.65696411132814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 810, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2997587710618973, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.015364667493849992, + "grad_norm": 0.10506005585193634, + "kept_token_count": 572.8453125, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9771534979343415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.924434951040894e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.03415412902832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2898223876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 820, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30021896958351135, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.015135047771036626, + "grad_norm": 0.0967249944806099, + "kept_token_count": 572.3890625, + "learning_rate": 1.658e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9771512567996978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.925302536226809e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.080641174316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.9055206298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 830, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.30022197365760805, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.017340341582894325, + "grad_norm": 0.1019851341843605, + "kept_token_count": 572.6328125, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9750173151493072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.758884625625797e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.48159408569336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.88037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 840, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2997033715248108, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.017514457739889622, + "grad_norm": 0.12219183146953583, + "kept_token_count": 572.978125, + "learning_rate": 1.698e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.974490511417389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.964658602257259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.850334548950194, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.10145263671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 850, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.29936706721782685, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.017196682281792164, + "grad_norm": 0.12111339718103409, + "kept_token_count": 573.5109375, + "learning_rate": 1.718e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.97537682056427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.61844481935259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.19268913269043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.4704528808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 860, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.300441375374794, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.017121990211308002, + "grad_norm": 0.11262472718954086, + "kept_token_count": 572.1875, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9737591743469238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010250340492348186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.96950569152832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.04710998535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 870, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.30084926784038546, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.017434647399932147, + "grad_norm": 0.09370590001344681, + "kept_token_count": 572.2859375, + "learning_rate": 1.758e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9741333246231079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010104185566888191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.581002426147464, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.11078186035155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 880, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.2995675057172775, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.01545078055933118, + "grad_norm": 0.10712796449661255, + "kept_token_count": 572.9703125, + "learning_rate": 1.778e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.976234632730484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.283366816816851e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.50009346008301, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.9817138671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 890, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2997311860322952, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.017597092408686878, + "grad_norm": 0.07481222599744797, + "kept_token_count": 572.9375, + "learning_rate": 1.798e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9763012826442719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.257327546947635e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.829783248901364, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3863922119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 900, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.3008278369903564, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.01634643180295825, + "grad_norm": 0.09579164534807205, + "kept_token_count": 571.9765625, + "learning_rate": 1.818e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9757313668727875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.47995540627744e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.383270263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4455291748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 910, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2990183889865875, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.014741571433842183, + "grad_norm": 0.11310672760009766, + "kept_token_count": 573.675, + "learning_rate": 1.838e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.976542842388153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.162974529317581e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.20060691833496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.0729156494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 920, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.3005524784326553, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.018028673063963652, + "grad_norm": 0.09959616512060165, + "kept_token_count": 572.44375, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9750715017318725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.73771428107284e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.760607147216795, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1806396484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 930, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2999930024147034, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.01642369357869029, + "grad_norm": 0.10042008012533188, + "kept_token_count": 572.7953125, + "learning_rate": 1.878e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9750348865985871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.75201757682953e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.86132125854492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.6561737060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 940, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2991655945777893, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.015545724239200354, + "grad_norm": 0.11070476472377777, + "kept_token_count": 573.7578125, + "learning_rate": 1.898e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9770670890808105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.958185062510893e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.60388031005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.55083923339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 950, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.3001699298620224, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.01672537988051772, + "grad_norm": 0.10919452458620071, + "kept_token_count": 572.8546875, + "learning_rate": 1.918e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.976213252544403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.291715832659975e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.89914665222168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.54909362792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 960, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.30047344863414766, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.01666149590164423, + "grad_norm": 0.12023551017045975, + "kept_token_count": 572.4, + "learning_rate": 1.938e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9743461072444916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010021070411312394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.1860107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6432281494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30032175183296206, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.013641040492802858, + "grad_norm": 0.08867582678794861, + "kept_token_count": 572.3921875, + "learning_rate": 1.958e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.97543283700943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.596574527677148e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.798051071166995, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2747833251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 980, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.299647119641304, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.020572414714843036, + "grad_norm": 0.13884863257408142, + "kept_token_count": 573.325, + "learning_rate": 1.978e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9741286754608154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010106002882821485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.636409759521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3025299072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 990, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.29908527731895446, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.015506533067673444, + "grad_norm": 0.12835198640823364, + "kept_token_count": 573.590625, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.974174028635025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010088292183354496, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.847921752929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7676300048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1000, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2991934448480606, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.01641427492722869, + "grad_norm": 0.12379823625087738, + "kept_token_count": 573.4515625, + "learning_rate": 2.018e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9742592096328735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010055018501589074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.08327941894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.2707580566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1010, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2997673094272614, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.020364236272871494, + "grad_norm": 0.15144410729408264, + "kept_token_count": 572.93125, + "learning_rate": 2.038e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9735471606254578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010333157988497987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.72063636779785, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.00291442871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3002774268388748, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.015957065112888812, + "grad_norm": 0.12930212914943695, + "kept_token_count": 572.6828125, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.016, + "mgd_cos_sim": 0.974921065568924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.796477825148031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.76444282531738, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 348.3883575439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1030, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.3000492453575134, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.01435534693300724, + "grad_norm": 0.12040257453918457, + "kept_token_count": 572.840625, + "learning_rate": 2.078e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9732285141944885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010457629250595346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.37306022644043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3885223388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1040, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3005062073469162, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.015690464805811644, + "grad_norm": 0.13519494235515594, + "kept_token_count": 572.6625, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9736461579799652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010294479652657173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.13865165710449, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.24037475585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1050, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.3000993669033051, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.013833693228662014, + "grad_norm": 0.12361705303192139, + "kept_token_count": 572.609375, + "learning_rate": 2.118e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.97460657954216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.919318617903627e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.786353302001956, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.42237243652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1060, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30054719746112823, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.016449503507465123, + "grad_norm": 0.11646189540624619, + "kept_token_count": 572.4671875, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9744177997112274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.993064231821336e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.450531005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.1844451904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1070, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.30058123767375944, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.01786943105980754, + "grad_norm": 0.1453218162059784, + "kept_token_count": 572.36875, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9740781188011169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0001012574888591189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.1098991394043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5770202636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1080, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2986455887556076, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.015103383269160987, + "grad_norm": 0.12331108748912811, + "kept_token_count": 573.875, + "learning_rate": 2.178e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.975572270154953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.542097832309082e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.085296630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.24810791015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1090, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.30060299932956697, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.015466985199600458, + "grad_norm": 0.09594842046499252, + "kept_token_count": 572.25625, + "learning_rate": 2.198e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.97549107670784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.5738116215216e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.134490203857425, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.6196807861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1100, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.3007912963628769, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.014528619684278965, + "grad_norm": 0.12848025560379028, + "kept_token_count": 572.178125, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.976576954126358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.149648394668475e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.412449264526366, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4527069091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1110, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2995819509029388, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.014710511546581984, + "grad_norm": 0.07776427268981934, + "kept_token_count": 573.234375, + "learning_rate": 2.238e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9770092487335205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.980779894045554e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.50868377685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.0588439941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1120, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29943064153194426, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.017561432812362908, + "grad_norm": 0.1815251260995865, + "kept_token_count": 573.340625, + "learning_rate": 2.258e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9769096553325654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.019687640829944e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.492950439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5537048339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1130, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.29998252391815183, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.017716331966221334, + "grad_norm": 0.14246779680252075, + "kept_token_count": 572.9796875, + "learning_rate": 2.278e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9741992056369781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010078454870381393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.63708992004395, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.19138488769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1140, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.30049416422843933, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.013968493323773146, + "grad_norm": 0.10992645472288132, + "kept_token_count": 572.2703125, + "learning_rate": 2.298e-05, + "loss": 0.014, + "mgd_cos_sim": 0.975463742017746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.584493891452439e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.99587783813477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.12681884765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1150, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29993900954723357, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.017755426932126284, + "grad_norm": 0.12690599262714386, + "kept_token_count": 572.8625, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9755121171474457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.56559058977291e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.35102157592773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.16700744628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1160, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30024024844169617, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.018821716029196977, + "grad_norm": 0.108591228723526, + "kept_token_count": 572.3640625, + "learning_rate": 2.338e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9749345123767853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.791225893422962e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.96727180480957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3976745605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1170, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.29921217262744904, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.017617226298898458, + "grad_norm": 0.12898391485214233, + "kept_token_count": 573.5515625, + "learning_rate": 2.358e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9736817896366119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0001028056321956683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.03928337097168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7057373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1180, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.29946932196617126, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.01487746424973011, + "grad_norm": 0.13654668629169464, + "kept_token_count": 573.2828125, + "learning_rate": 2.378e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9755077719688415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.567293163854629e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.85676536560059, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8372406005859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1190, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2987325370311737, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.018424554727971554, + "grad_norm": 0.15535050630569458, + "kept_token_count": 573.9765625, + "learning_rate": 2.398e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9748456478118896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.825941160670481e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.73169364929199, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6744354248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1200, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.2996218532323837, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.015084878914058209, + "grad_norm": 0.1055995374917984, + "kept_token_count": 572.85, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9755310237407684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.558213278069161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.55749092102051, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.36510009765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1210, + "valid_token_count": 817.9078125 + }, + { + "actual_token_mask_ratio": 0.3002124309539795, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.015160552319139242, + "grad_norm": 0.1399277001619339, + "kept_token_count": 572.6015625, + "learning_rate": 2.438e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9750585734844208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.742761685629375e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.998114013671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6625091552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1220, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2996974796056747, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.017289489693939685, + "grad_norm": 0.16123060882091522, + "kept_token_count": 572.6765625, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9749779880046845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.774238060344942e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.805165481567386, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2597961425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1230, + "valid_token_count": 817.75625 + }, + { + "actual_token_mask_ratio": 0.3001063078641891, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.016097676102072002, + "grad_norm": 0.12836724519729614, + "kept_token_count": 572.690625, + "learning_rate": 2.478e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9747665047645568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.856847318587825e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.28768768310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6551208496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1240, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30077705085277556, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.016619077976793052, + "grad_norm": 0.1662178784608841, + "kept_token_count": 572.090625, + "learning_rate": 2.498e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9757164239883422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.485790869803167e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.25268898010254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5819610595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1250, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2985852837562561, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.016675494704395534, + "grad_norm": 0.11897339671850204, + "kept_token_count": 574.0421875, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9770059466362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.982069630292244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.6240535736084, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.06662902832034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1260, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.29966589212417605, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.0153251969255507, + "grad_norm": 0.116988904774189, + "kept_token_count": 573.03125, + "learning_rate": 2.538e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9758484482765197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.434218372916803e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.86366958618164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.09739074707034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1270, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3007175445556641, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.017523884773254395, + "grad_norm": 0.11847742646932602, + "kept_token_count": 572.215625, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9769139289855957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.018019190989434e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.59543647766113, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.6775299072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1280, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2998242437839508, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.016528843343257903, + "grad_norm": 0.11035176366567612, + "kept_token_count": 572.68125, + "learning_rate": 2.578e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9763991177082062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.219109051628038e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.305349349975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.84380798339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1290, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.30048237442970277, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.01652304222807288, + "grad_norm": 0.12241700291633606, + "kept_token_count": 572.2765625, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.976138687133789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.320839817519299e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.70455131530762, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.02495727539065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1300, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.300679674744606, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.0148368822876364, + "grad_norm": 0.10264571756124496, + "kept_token_count": 572.2796875, + "learning_rate": 2.618e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9754787504673004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.578633034834638e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.523193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2510009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.299919393658638, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.014127597771584987, + "grad_norm": 0.1183651015162468, + "kept_token_count": 572.8375, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9766635537147522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.115820721490308e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.52941780090332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.5089630126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1320, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29961993992328645, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.019325973372906445, + "grad_norm": 0.12955713272094727, + "kept_token_count": 573.21875, + "learning_rate": 2.658e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9752459824085236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.669553037383594e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.02932510375977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.57847900390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1330, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.3013006836175919, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.018098668847233056, + "grad_norm": 0.15305384993553162, + "kept_token_count": 571.9515625, + "learning_rate": 2.678e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9750347077846527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.75208604359068e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.72497634887695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.51856994628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1340, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.3002269685268402, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.0162500879727304, + "grad_norm": 0.115387462079525, + "kept_token_count": 572.6109375, + "learning_rate": 2.698e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9750451862812042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.747990989126266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.218048095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.26575927734376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1350, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29934667348861693, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.016548411827534436, + "grad_norm": 0.09851456433534622, + "kept_token_count": 573.2921875, + "learning_rate": 2.718e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9745606243610382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.937273061950691e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.916372299194336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.00348510742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1360, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3000524193048477, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.017910389229655266, + "grad_norm": 0.11803051829338074, + "kept_token_count": 572.840625, + "learning_rate": 2.738e-05, + "loss": 0.018, + "mgd_cos_sim": 0.973993718624115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010158722798223608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.52316551208496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.33250732421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1370, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3007178455591202, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.01530330004170537, + "grad_norm": 0.09434033185243607, + "kept_token_count": 572.3640625, + "learning_rate": 2.758e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.976328557729721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.24667896470055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.7059871673584, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.16444702148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1380, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.299382883310318, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.016456866543740033, + "grad_norm": 0.10515587776899338, + "kept_token_count": 573.2015625, + "learning_rate": 2.778e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9750062704086304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.763193957041949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.30475692749023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.3406555175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1390, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29928942024707794, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.01573877679184079, + "grad_norm": 0.130801260471344, + "kept_token_count": 573.53125, + "learning_rate": 2.798e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9766312062740325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.1284497466404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.00409202575683, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4602844238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1400, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2998581796884537, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.01589431203901768, + "grad_norm": 0.14398962259292603, + "kept_token_count": 572.959375, + "learning_rate": 2.818e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9747964262962341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.845164458965883e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.576330947875974, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.81528625488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1410, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.29935629665851593, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.01700674956664443, + "grad_norm": 0.08988562226295471, + "kept_token_count": 573.24375, + "learning_rate": 2.838e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9744880557060241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.965616118279286e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.80805778503418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3811798095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1420, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29961307942867277, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.01590289520099759, + "grad_norm": 0.09494459629058838, + "kept_token_count": 573.0328125, + "learning_rate": 2.858e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9759685039520264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.387319514644332e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.20617828369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.855908203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.300286078453064, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.015000889450311661, + "grad_norm": 0.10662087053060532, + "kept_token_count": 572.4859375, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9760284543037414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.363899807794951e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.38585662841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2454467773438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1440, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.3006762474775314, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.015314933843910694, + "grad_norm": 0.11985492706298828, + "kept_token_count": 572.275, + "learning_rate": 2.898e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9761963129043579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.298339573433622e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.070217514038085, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6929626464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1450, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3003896355628967, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.020711323525756598, + "grad_norm": 0.1318219155073166, + "kept_token_count": 572.359375, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.974746149778366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.864802923402749e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.259457397460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.44135131835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1460, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.3003167450428009, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.016637415904551745, + "grad_norm": 0.09212473034858704, + "kept_token_count": 572.56875, + "learning_rate": 2.938e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9762449026107788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.279357618652284e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.4676456451416, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2914794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1470, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2999399185180664, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.018136495258659124, + "grad_norm": 0.11387897282838821, + "kept_token_count": 572.98125, + "learning_rate": 2.958e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9758665561676025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.427143668290227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.6911003112793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.25348510742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1480, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2991121977567673, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.017118398752063512, + "grad_norm": 0.16151855885982513, + "kept_token_count": 573.4875, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9752124130725861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.68267566349823e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.89088478088379, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.918994140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1490, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3000966399908066, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.01627791440114379, + "grad_norm": 0.1530475914478302, + "kept_token_count": 572.8421875, + "learning_rate": 2.998e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9744012355804443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.999536268878728e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.45909118652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.07269287109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1500, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.3003915578126907, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.017006177641451358, + "grad_norm": 0.11578233540058136, + "kept_token_count": 572.1484375, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9740055620670318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010154098854400217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.17376441955567, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.53049621582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1510, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.29991304874420166, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.014818290900439023, + "grad_norm": 0.1188674047589302, + "kept_token_count": 572.7828125, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9736298143863678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.000103008774749469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.74097175598145, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9940246582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1520, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30014065504074094, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.017134217265993356, + "grad_norm": 0.10621608048677444, + "kept_token_count": 572.8421875, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9770139157772064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.97896330570802e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.349129104614256, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5850494384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1530, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.29975709319114685, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.017628259025514126, + "grad_norm": 0.12923985719680786, + "kept_token_count": 573.0328125, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9753275156021118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.637704570195638e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.65357284545898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.64707946777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1540, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3004143059253693, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.01782621946185827, + "grad_norm": 0.12135828286409378, + "kept_token_count": 572.771875, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.975907689332962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.411077189724892e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.74061126708985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5086242675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1550, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.30034820437431337, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.01845763921737671, + "grad_norm": 0.0929538682103157, + "kept_token_count": 572.35625, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9754262745380402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.599134064046666e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.88346366882324, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.53907775878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1560, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.3000228315591812, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.018725086748600007, + "grad_norm": 0.1044842079281807, + "kept_token_count": 572.8234375, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9757854759693145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.458817075937986e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.26457405090332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.42608642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1570, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2993821054697037, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.016895493026822805, + "grad_norm": 0.13825687766075134, + "kept_token_count": 573.140625, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9742307066917419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010066145696328022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.988690567016604, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8774139404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1580, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.3001670718193054, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.017401625588536262, + "grad_norm": 0.10954520106315613, + "kept_token_count": 572.8078125, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9741547286510468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.000100958306575194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.014990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9404998779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1590, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.29929962158203127, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.018162289541214703, + "grad_norm": 0.1028742641210556, + "kept_token_count": 573.425, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9738576829433441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010211863336735405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.30899238586426, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.95416259765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1600, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30020821988582613, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.017822423577308656, + "grad_norm": 0.09617789834737778, + "kept_token_count": 572.60625, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9732902407646179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010433521601953544, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.362820053100585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9650451660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1610, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.2997290253639221, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.01568181496113539, + "grad_norm": 0.0882333517074585, + "kept_token_count": 572.75625, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9740807414054871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010124729669769295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.21348342895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5177459716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1620, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.3005290776491165, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.01989307887852192, + "grad_norm": 0.1093216985464096, + "kept_token_count": 572.209375, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9744391858577728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.984708012780174e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.4718204498291, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.695263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3006167560815811, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.01552248215302825, + "grad_norm": 0.14315654337406158, + "kept_token_count": 572.3515625, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9738332331180573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010221415141131729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.4685474395752, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.5214416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.30047534704208373, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.016649116948246955, + "grad_norm": 0.11331919580698013, + "kept_token_count": 572.309375, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9745966851711273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.923181351041421e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.89818992614746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.95575256347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2996360182762146, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.01546188979409635, + "grad_norm": 0.12796548008918762, + "kept_token_count": 573.0671875, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9768280029296875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.051576053025201e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.26332054138183, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.45705261230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1660, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.301080322265625, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.017558668181300162, + "grad_norm": 0.10491295903921127, + "kept_token_count": 572.1515625, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9765811026096344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.148027311312035e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.36646690368653, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.95531005859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1670, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.2988768994808197, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.018121974356472493, + "grad_norm": 0.1351565271615982, + "kept_token_count": 573.7578125, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9761955678462982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.298626318923197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.73095092773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.20863647460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1680, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.3003523677587509, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.019951824657619, + "grad_norm": 0.14334224164485931, + "kept_token_count": 572.3984375, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9761442422866822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.318677039118483e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.05204315185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4154449462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.29947859346866607, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.020182935055345297, + "grad_norm": 0.11945340782403946, + "kept_token_count": 573.2515625, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.976906806230545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.020791476359591e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.14253044128418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.61838989257814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2992993235588074, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.017456484865397214, + "grad_norm": 0.13003475964069366, + "kept_token_count": 573.334375, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9771203875541687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.937366437749006e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.977423095703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7225402832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1710, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29997639656066893, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.017669288162142038, + "grad_norm": 0.13445761799812317, + "kept_token_count": 572.6984375, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9768027782440185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.061424934770912e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.1668312072754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.60569458007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1720, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2999011218547821, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.017049440834671258, + "grad_norm": 0.12219177931547165, + "kept_token_count": 572.796875, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9760575473308564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.352538399980404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.369673538208, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.98997497558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1730, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.30005930066108705, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.017335285525768997, + "grad_norm": 0.1094033345580101, + "kept_token_count": 572.75, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.974983412027359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.772129851626232e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.49313812255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7216003417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1740, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2995916843414307, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.018647576123476027, + "grad_norm": 0.1309262216091156, + "kept_token_count": 573.1296875, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9734359622001648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010376599384471774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.110589599609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.582861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1750, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30057517886161805, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.01623215340077877, + "grad_norm": 0.1272682398557663, + "kept_token_count": 572.4421875, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9764125883579254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.213845842168667e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.88941421508789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3909606933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1760, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.3005957692861557, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.019126639422029256, + "grad_norm": 0.15107600390911102, + "kept_token_count": 572.0796875, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.975398713350296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.609898916096427e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.848649978637695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0730010986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1770, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.2997490853071213, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.016737076081335546, + "grad_norm": 0.10729068517684937, + "kept_token_count": 573.028125, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9758999168872833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.414113010279835e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.23893432617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.95961303710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1780, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3000136911869049, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.018520320858806373, + "grad_norm": 0.16956529021263123, + "kept_token_count": 572.7046875, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9766422748565674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.124131975113414e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.11561317443848, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.3845458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1790, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.30009452998638153, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.018281765282154083, + "grad_norm": 0.14927799999713898, + "kept_token_count": 572.715625, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.975033450126648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.75258146354463e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.525489807128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5484313964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1800, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3002817243337631, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.016214008536189796, + "grad_norm": 0.1301700621843338, + "kept_token_count": 572.4859375, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9753160417079926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.642187505960464e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.452642822265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.598583984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1810, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30030458569526675, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.017846724297851325, + "grad_norm": 0.11209028959274292, + "kept_token_count": 572.5515625, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.974481600522995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.968143422156572e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.69965591430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.93421630859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1820, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29960410892963407, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.016162376292049883, + "grad_norm": 0.12252196669578552, + "kept_token_count": 573.36875, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9740685522556305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010129493530257605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.608590316772464, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5329559326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1830, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.30037048161029817, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.019582550134509803, + "grad_norm": 0.12627334892749786, + "kept_token_count": 572.553125, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9739888608455658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0001016062575217802, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.58748893737793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.63392944335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1840, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2998535603284836, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.017825637478381395, + "grad_norm": 0.14231926202774048, + "kept_token_count": 572.7546875, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9749194800853729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.797094244277105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.59606018066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.79852600097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1850, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.3000787734985352, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.018986041098833083, + "grad_norm": 0.1174439787864685, + "kept_token_count": 572.571875, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9751452684402466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.708896322990768e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.99049682617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2914245605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1860, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.300685653090477, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.01751186829060316, + "grad_norm": 0.09410379827022552, + "kept_token_count": 572.2671875, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9765727639198303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.151279155048542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.206795120239256, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.0762115478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1870, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.30138199031352997, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.016751452535390853, + "grad_norm": 0.12882044911384583, + "kept_token_count": 571.7640625, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9771061480045319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.942927452153527e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.0890106201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4084991455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1880, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2996022790670395, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.019187751691788435, + "grad_norm": 0.11434072256088257, + "kept_token_count": 573.271875, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9757122993469238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.487401039223186e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.518714904785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9483245849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1890, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.2990962713956833, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.017239136341959237, + "grad_norm": 0.15738511085510254, + "kept_token_count": 573.5921875, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9739309310913086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010183249032706954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.69880981445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.11339416503904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1900, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.29937402009963987, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.018213983718305828, + "grad_norm": 0.1309971958398819, + "kept_token_count": 573.3203125, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9747850954532623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.849593989201822e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.23790473937989, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1913116455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29988237619400027, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.016954691801220178, + "grad_norm": 0.11585217714309692, + "kept_token_count": 572.7625, + "learning_rate": 2.998400360465264e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9752443492412567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.670196886872873e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.49836769104004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.35328369140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.30016522407531737, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.01767076728865504, + "grad_norm": 0.11130496859550476, + "kept_token_count": 572.415625, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9756254434585572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.521327228867449e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.13218460083008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6027404785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1930, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.3000144213438034, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.018979059439152478, + "grad_norm": 0.10345816612243652, + "kept_token_count": 572.7890625, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9762535810470581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.275970442104153e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.50927467346192, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.5858428955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2990323483943939, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.01717514041811228, + "grad_norm": 0.13590215146541595, + "kept_token_count": 573.575, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9767888784408569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.06686022062786e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.02381286621093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5391876220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1950, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2990981459617615, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.01650518449023366, + "grad_norm": 0.11251315474510193, + "kept_token_count": 573.58125, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9785891473293304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.363630331587046e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.42369537353515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.51015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1960, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3010513842105865, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.0195637340657413, + "grad_norm": 0.17341835796833038, + "kept_token_count": 571.9609375, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9766495406627655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.121292387135327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.45133666992187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.50611572265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1970, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2996784269809723, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.01811526417732239, + "grad_norm": 0.1514906883239746, + "kept_token_count": 573.2140625, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9756536722183228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.5103083003778e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.541819381713864, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.10528869628905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1980, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2992655009031296, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.019901890587061644, + "grad_norm": 0.10494184494018555, + "kept_token_count": 573.41875, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9741214752197266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010108817150467075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.680607604980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4020080566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1990, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2997048169374466, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.016374375484883786, + "grad_norm": 0.10928941518068314, + "kept_token_count": 573.028125, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9764270603656768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.208197297994047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.3371208190918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.88886108398435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2000, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3001808226108551, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.01780729591846466, + "grad_norm": 0.14623893797397614, + "kept_token_count": 572.4875, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9751605093479156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.702946408651769e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.51329231262207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7809783935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.3009602248668671, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.01894253082573414, + "grad_norm": 0.12012264877557755, + "kept_token_count": 571.9484375, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.019, + "mgd_cos_sim": 0.977267324924469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.879975721356458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.98385162353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6117401123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2020, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2994983226060867, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.01868387833237648, + "grad_norm": 0.16416603326797485, + "kept_token_count": 573.0984375, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9753840446472168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.615632297936827e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.54904365539551, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.13368225097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2030, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2993212819099426, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.016891192086040972, + "grad_norm": 0.10346546769142151, + "kept_token_count": 573.2640625, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9758216738700867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.444677125429735e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.4530532836914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1980285644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2040, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3001344382762909, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.019123245775699616, + "grad_norm": 0.14748841524124146, + "kept_token_count": 572.6640625, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9749091625213623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.801129926927388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.81921195983887, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.69411315917966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30033210217952727, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.018317315354943277, + "grad_norm": 0.12539778649806976, + "kept_token_count": 572.5140625, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9754791378974914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.578479075571522e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.215702819824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5791534423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2060, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3002815693616867, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.020553888007998466, + "grad_norm": 0.12089744955301285, + "kept_token_count": 572.446875, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9724339485168457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010768005377030932, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.785617446899415, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3118377685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2070, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.29957151114940644, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.016733760107308628, + "grad_norm": 0.15090015530586243, + "kept_token_count": 573.1859375, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9745018601417541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.960231982404366e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.93489456176758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8915588378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2080, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.298609060049057, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.01877921922132373, + "grad_norm": 0.1292172074317932, + "kept_token_count": 573.7328125, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9763607084751129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.234115495928563e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.32768630981445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.12481689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2090, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29884508848190305, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.018075068946927786, + "grad_norm": 0.13735522329807281, + "kept_token_count": 573.8703125, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9748060464859009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.841409846558235e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.7685920715332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.51332702636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2100, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.3008542835712433, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.015972763858735563, + "grad_norm": 0.12730377912521362, + "kept_token_count": 572.146875, + "learning_rate": 2.996621353380638e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9743083715438843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010035806117230095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.748311614990236, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.15972290039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2110, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.301345694065094, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.017443985119462012, + "grad_norm": 0.15154407918453217, + "kept_token_count": 571.75, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9764589488506317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.195740422001109e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.28475036621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.85478515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2120, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.3000164061784744, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.0205965805798769, + "grad_norm": 0.10422714054584503, + "kept_token_count": 572.90625, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9733651578426361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010404250278952531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.07295265197754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01905822753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2130, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.3005957961082458, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.01612067800015211, + "grad_norm": 0.12512220442295074, + "kept_token_count": 572.171875, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9754603028297424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.585838924976997e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.860599899292, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8644653320313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2993177860975266, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.016896069142967464, + "grad_norm": 0.109325110912323, + "kept_token_count": 573.46875, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9764479279518128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.200044660246931e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.890282821655276, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.8715576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2150, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.3016852766275406, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.016999358497560024, + "grad_norm": 0.11239714175462723, + "kept_token_count": 571.4890625, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9765235126018524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.17051889700815e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.40828666687012, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.75872192382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2160, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2999396026134491, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.015118560846894979, + "grad_norm": 0.1498004049062729, + "kept_token_count": 572.7265625, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.976785945892334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.06800625671167e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.71203689575195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8765838623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2170, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3000690460205078, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.017654489446431398, + "grad_norm": 0.1207231655716896, + "kept_token_count": 572.7984375, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9741881728172302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010082760782097467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.58624534606933, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9989868164063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2180, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3003836303949356, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.01623279843479395, + "grad_norm": 0.10871193557977676, + "kept_token_count": 572.328125, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9753053426742554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.646368562243879e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.569136428833005, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.43812561035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.30086955726146697, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.021013366896659134, + "grad_norm": 0.1739395558834076, + "kept_token_count": 572.2625, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9753204166889191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.640481730457395e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.03533401489258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.41497802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2200, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.2994951128959656, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.019229062274098395, + "grad_norm": 0.14397253096103668, + "kept_token_count": 573.4296875, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9758775353431701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.422855655429884e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.27687797546386, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5727813720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2210, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.29997539818286895, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.0192116130143404, + "grad_norm": 0.1342276781797409, + "kept_token_count": 572.93125, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9765157639980316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.173545404337347e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.75162048339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.2506958007813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2220, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.30060167610645294, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.017199407704174517, + "grad_norm": 0.11900472640991211, + "kept_token_count": 572.453125, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9757898330688477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.457115302211606e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.24141235351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9531219482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2230, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.3001558780670166, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.016927321907132864, + "grad_norm": 0.12462559342384338, + "kept_token_count": 572.81875, + "learning_rate": 2.995025837242257e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9770283997058868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.973300427896901e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.4895565032959, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.59755859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2240, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.29998689591884614, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.018290670216083528, + "grad_norm": 0.11633387953042984, + "kept_token_count": 572.79375, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9760581314563751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.352308916277252e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.13779487609864, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1668182373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.30070303976535795, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.018778700567781925, + "grad_norm": 0.10009339451789856, + "kept_token_count": 572.196875, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9775049149990082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.787164624664001e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.89361038208008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.22618713378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2260, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3005128264427185, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.017735042329877614, + "grad_norm": 0.09972306340932846, + "kept_token_count": 572.24375, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9762872874736785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.262801104341634e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.9364013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4681732177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2270, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.30063513815402987, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.015157163701951503, + "grad_norm": 0.10896338522434235, + "kept_token_count": 572.2609375, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9756976306438446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.493126417510211e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.7746425628662, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4576873779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2280, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.3003144204616547, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.017870585899800062, + "grad_norm": 0.09879978746175766, + "kept_token_count": 572.4828125, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9751782596111298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.696009437902831e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.21905517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.43626098632814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30071198046207426, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.01757050585001707, + "grad_norm": 0.13351218402385712, + "kept_token_count": 571.9765625, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9753009676933289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.64807957643643e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.26786994934082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.79990844726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2300, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.30003742277622225, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.019216987676918505, + "grad_norm": 0.17887970805168152, + "kept_token_count": 572.978125, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9759907245635986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.378646791446954e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.58972854614258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1769592285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2310, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.30049399137496946, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.018307303451001643, + "grad_norm": 0.1476134955883026, + "kept_token_count": 572.3734375, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9766008496284485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.140311231021769e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.55630035400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.1275207519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29964849054813386, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.019980833120644092, + "grad_norm": 0.10518749058246613, + "kept_token_count": 573.0203125, + "learning_rate": 2.993741386831634e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9767625093460083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.077162103494629e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.7253288269043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0500183105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2330, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2998682111501694, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.01696718968451023, + "grad_norm": 0.11111649125814438, + "kept_token_count": 573.059375, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9750479221343994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.74692273302935e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.09350929260253, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2887268066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2987920939922333, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.020841535367071628, + "grad_norm": 0.10617823898792267, + "kept_token_count": 573.6140625, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9755124568939209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.565465661580675e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.92167663574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.75111694335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2350, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.2993723154067993, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.019021239317953586, + "grad_norm": 0.1707143783569336, + "kept_token_count": 573.290625, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9759683609008789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.387377212988213e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.68855247497558, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9176513671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30045910775661466, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.01793078100308776, + "grad_norm": 0.11300373077392578, + "kept_token_count": 572.403125, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9754602789878846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.585853986209259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.2294189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.08494873046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2370, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30024502575397494, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.017327898368239402, + "grad_norm": 0.15394803881645203, + "kept_token_count": 572.6765625, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9760118901729584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.370375846629031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.19540252685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.34092102050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2380, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2996518164873123, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.018516109324991703, + "grad_norm": 0.1415599137544632, + "kept_token_count": 573.3703125, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9763117611408234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.253238313249313e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.20533866882325, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.72515563964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2390, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.29979282021522524, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.01921751918271184, + "grad_norm": 0.13578689098358154, + "kept_token_count": 573.1734375, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9754974246025085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.571340924594552e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.894616317749026, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.774658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2400, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.3003216922283173, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.01611714381724596, + "grad_norm": 0.13011179864406586, + "kept_token_count": 572.5578125, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9761978268623352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.297737560700625e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.09270896911622, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6385559082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2410, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2996482402086258, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.02184778666123748, + "grad_norm": 0.10073450952768326, + "kept_token_count": 572.9296875, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9762639582157135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.271908274968154e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.90841445922852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.005029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2420, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.29982700645923616, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.018943377397954465, + "grad_norm": 0.14515051245689392, + "kept_token_count": 572.7453125, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9747049927711486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.880883808364161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.17955322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1593811035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2430, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.3000150263309479, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.01947758188471198, + "grad_norm": 0.12471365928649902, + "kept_token_count": 572.9203125, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9753666520118713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.622419383958913e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.02141532897949, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.7544830322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2440, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.3004046559333801, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.018935344088822602, + "grad_norm": 0.10450557619333267, + "kept_token_count": 572.659375, + "learning_rate": 2.991800117455e-05, + "loss": 0.019, + "mgd_cos_sim": 0.974461829662323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.975861248676666e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.11562004089355, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.04954528808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2450, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3008372336626053, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.017139494977891447, + "grad_norm": 0.10786442458629608, + "kept_token_count": 572.4546875, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9740113914012909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010151820242754184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.8468734741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.47760009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2460, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.2997002422809601, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.019188872817903758, + "grad_norm": 0.12893754243850708, + "kept_token_count": 573.4, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9751386284828186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.71149216638878e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.53141975402832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.74267578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2470, + "valid_token_count": 818.7953125 + }, + { + "actual_token_mask_ratio": 0.2991443336009979, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.016590028535574675, + "grad_norm": 0.10071729868650436, + "kept_token_count": 573.5078125, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9750286936759949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.754438142408616e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.09000205993652, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.16590270996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2480, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3005495220422745, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.020716474670916796, + "grad_norm": 0.13130724430084229, + "kept_token_count": 572.478125, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9766288936138153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.129360259976237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.7182403564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 349.95342407226565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2490, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.30017416179180145, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.01874116025865078, + "grad_norm": 0.1148001030087471, + "kept_token_count": 572.65625, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9745678305625916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.934460904332809e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.34141426086425, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1797637939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2500, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2997656911611557, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.01862995224073529, + "grad_norm": 0.11056114733219147, + "kept_token_count": 573.15625, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9749450504779815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.787106682779267e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.19257888793945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.49456481933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2510, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2994685560464859, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.018351499550044538, + "grad_norm": 0.12195567041635513, + "kept_token_count": 573.1546875, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9751618385314942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.702425304567442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.11115341186523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.0486633300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2989888101816177, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.017320245131850244, + "grad_norm": 0.09483745694160461, + "kept_token_count": 573.6890625, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.976580286026001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.14834403374698e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.94838981628418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3451782226563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2530, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.30026874542236326, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.017616932280361652, + "grad_norm": 0.15275998413562775, + "kept_token_count": 572.4453125, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9760731875896453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.346429142169655e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.39747352600098, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1727020263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2540, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.29996303021907805, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.01947700176388025, + "grad_norm": 0.10910488665103912, + "kept_token_count": 572.5953125, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9743947744369507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010002059934777208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.22672348022461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6594512939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2550, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.30077953934669494, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.01975620910525322, + "grad_norm": 0.13001690804958344, + "kept_token_count": 572.1640625, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9737764656543731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010243592550978065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.56107521057129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4454010009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2560, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29933708608150483, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.017006369680166243, + "grad_norm": 0.14430098235607147, + "kept_token_count": 573.4640625, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9754914224147797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.573680945322849e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.04595184326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.05027465820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2570, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.2997824549674988, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.01902609383687377, + "grad_norm": 0.10320524871349335, + "kept_token_count": 573.1609375, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9755328834056854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.557483062963001e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.84198455810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.5930603027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2580, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.3000083327293396, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.015046012867242099, + "grad_norm": 0.09381376206874847, + "kept_token_count": 572.71875, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9741278469562531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010106328627443873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.49228477478027, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.7556579589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2590, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.29967465698719026, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.020095604471862316, + "grad_norm": 0.12278646230697632, + "kept_token_count": 573.096875, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9747865617275238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.849022317212074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.610944747924805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4262451171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2600, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3005633920431137, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.0189189825206995, + "grad_norm": 0.11717665940523148, + "kept_token_count": 572.284375, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9750624656677246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.741240864968859e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.5203742980957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.99652404785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2610, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.30017467141151427, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.01871516928076744, + "grad_norm": 0.13873252272605896, + "kept_token_count": 572.44375, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9754400312900543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.593755021342076e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.25014114379883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.2362060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2620, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.3001000702381134, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.017614104319363832, + "grad_norm": 0.11581792682409286, + "kept_token_count": 572.659375, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9762226104736328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.288064829888753e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.93999633789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9163452148438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2630, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2991839170455933, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.019015376456081867, + "grad_norm": 0.12458382546901703, + "kept_token_count": 573.696875, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9753323912620544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.635803871788085e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.93875389099121, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.3255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2640, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.3010301887989044, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.01894203815609217, + "grad_norm": 0.14450043439865112, + "kept_token_count": 571.8671875, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.019, + "mgd_cos_sim": 0.975211638212204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.68296910286881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.77664184570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.49135437011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3006019562482834, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.018684784695506097, + "grad_norm": 0.1415587216615677, + "kept_token_count": 572.21875, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9745728611946106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.932503526215442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.32354164123535, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2519958496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2660, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.3000429689884186, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.01891694599762559, + "grad_norm": 0.1360117644071579, + "kept_token_count": 572.5796875, + "learning_rate": 2.987563468543624e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9753279328346253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.637543116696179e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.12862014770508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.583837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2670, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.2995481312274933, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.020355944521725178, + "grad_norm": 0.15484990179538727, + "kept_token_count": 573.13125, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9766385674476623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.125581564148889e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.01269912719727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31023864746095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2680, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29917662739753725, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.01904274271801114, + "grad_norm": 0.12864093482494354, + "kept_token_count": 573.61875, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9770473539829254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.965895103756339e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.18727416992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.4726806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2690, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.3002978593111038, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.01926359673961997, + "grad_norm": 0.13402505218982697, + "kept_token_count": 572.35625, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9752066016197205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.684940159786493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.73253021240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7212310791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2700, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.29885373413562777, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.017456672340631484, + "grad_norm": 0.146734818816185, + "kept_token_count": 573.8109375, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9755633354187012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.545594875817187e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.452158737182614, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5430053710937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2710, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.298947674036026, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.01840018145740032, + "grad_norm": 0.17046648263931274, + "kept_token_count": 573.51875, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9756185412406921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.524024353595451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.58835525512696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6075042724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.299894180893898, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.017991083301603795, + "grad_norm": 0.15353810787200928, + "kept_token_count": 572.8609375, + "learning_rate": 2.986256076618961e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9760423719882965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.358467068523168e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.41024627685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.39549560546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2730, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2991007030010223, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.016870256513357162, + "grad_norm": 0.14904119074344635, + "kept_token_count": 573.3125, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9765784323215485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.14906595426146e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.51215934753418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1232421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2740, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.3006491720676422, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.017175741586834194, + "grad_norm": 0.1432710736989975, + "kept_token_count": 572.1171875, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9763472735881805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.2393684462877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.88922653198242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2466613769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2750, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.2995342701673508, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.016881045140326022, + "grad_norm": 0.12258755415678024, + "kept_token_count": 573.2515625, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9759100794792175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.410147031303495e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.80590362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4610870361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2760, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29896354079246523, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.020882341545075178, + "grad_norm": 0.13530398905277252, + "kept_token_count": 573.3828125, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9754126608371735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.604446313460357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.92427520751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.706787109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2770, + "valid_token_count": 817.9015625 + }, + { + "actual_token_mask_ratio": 0.3004766762256622, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.017836169712245463, + "grad_norm": 0.16591861844062805, + "kept_token_count": 572.084375, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9756635189056396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.506459391559475e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.0583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8020385742187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2780, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.2989380359649658, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.01808167975395918, + "grad_norm": 0.12567929923534393, + "kept_token_count": 573.553125, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9755018413066864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.569598259986378e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.95968017578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.85535583496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2989553540945053, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.01931262919679284, + "grad_norm": 0.18650847673416138, + "kept_token_count": 573.4234375, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9769313275814057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.011219299281947e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.07735977172851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 351.8615783691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2800, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.30098783075809477, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.021204858645796777, + "grad_norm": 0.1412995308637619, + "kept_token_count": 572.1203125, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9741535604000091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010096283149323426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.50734748840332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7454772949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2810, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.30126091539859773, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.019181032292544843, + "grad_norm": 0.11646334081888199, + "kept_token_count": 571.8140625, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9739729583263397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010166824140469543, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.83552284240723, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5909484863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2820, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2996456414461136, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.016388476919382812, + "grad_norm": 0.11283309757709503, + "kept_token_count": 573.2546875, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9765664756298065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.15373741008807e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.58374557495117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.18323974609376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2830, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.30025793313980104, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.016603822261095046, + "grad_norm": 0.09741091728210449, + "kept_token_count": 572.6046875, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9762952208518982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.259695871151052e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.3703239440918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.4459930419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2840, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2995491623878479, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.01738253068178892, + "grad_norm": 0.11852115392684937, + "kept_token_count": 573.2078125, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9763343691825866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.24440595554188e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.04236755371093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8990814208984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2850, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.299423810839653, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.020811316464096306, + "grad_norm": 0.1173284575343132, + "kept_token_count": 573.43125, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9777977168560028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.672788535477594e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.6865020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4989013671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2860, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.2994837015867233, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.019246276654303073, + "grad_norm": 0.1121484711766243, + "kept_token_count": 573.05625, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9773590683937072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.844126859912649e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.9178565979004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0176666259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2870, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.29976303279399874, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.01849098550155759, + "grad_norm": 0.11150628328323364, + "kept_token_count": 572.8765625, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.975584876537323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.537175428704358e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.15923080444335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2749938964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2880, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.29971725046634673, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.020147091895341872, + "grad_norm": 0.12210965901613235, + "kept_token_count": 573.021875, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.975185364484787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.693235988379456e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.91457824707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0135192871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2890, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2990907222032547, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.01778542660176754, + "grad_norm": 0.11485997587442398, + "kept_token_count": 573.3453125, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9770086169242859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.981028367998079e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.44933395385742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.4786865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2900, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.29930259883403776, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.017201646231114864, + "grad_norm": 0.14261990785598755, + "kept_token_count": 573.528125, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9772506177425384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.886493305908516e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.11322479248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5102264404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2910, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.300349822640419, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.020821194909512997, + "grad_norm": 0.11740534752607346, + "kept_token_count": 572.3734375, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9752775192260742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.6572488109814e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.96300430297852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4051086425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2993274986743927, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.01896490566432476, + "grad_norm": 0.13494238257408142, + "kept_token_count": 573.5921875, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9759831309318543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.38160934310872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.45364379882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5285583496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2930, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.29993259310722353, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.02147962572053075, + "grad_norm": 0.15133608877658844, + "kept_token_count": 572.89375, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9761797070503235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.30481961404439e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.7670051574707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3893585205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2940, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29983174204826357, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.017625757679343223, + "grad_norm": 0.14395536482334137, + "kept_token_count": 573.14375, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9752217888832092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.679010181571357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.6087760925293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6078582763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2950, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.29947091042995455, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.01911517223343253, + "grad_norm": 0.12874923646450043, + "kept_token_count": 573.3546875, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9754218280315399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.600863122614101e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.14043502807617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7557067871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2960, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.299652436375618, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.017405567411333322, + "grad_norm": 0.12525972723960876, + "kept_token_count": 573.075, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9751480400562287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.707817735034041e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.85341911315918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6299102783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.29958401918411254, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.01782976696267724, + "grad_norm": 0.09761440008878708, + "kept_token_count": 573.23125, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9745553135871887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.939350202330388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.97311859130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4185821533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.2986153572797775, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.016100457031279804, + "grad_norm": 0.16830742359161377, + "kept_token_count": 573.740625, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9759103775024414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.410024285898544e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.21729736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2969116210937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2990, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.30004737377166746, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.016804720275104046, + "grad_norm": 0.11665590107440948, + "kept_token_count": 573.053125, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9760258734226227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.36491254833527e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.68333435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.40549621582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3000, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.30023394227027894, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.01763057690113783, + "grad_norm": 0.1737852394580841, + "kept_token_count": 572.6765625, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9755799829959869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.539087186567485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.70796813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3916809082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3010, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3000565230846405, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.018815619871020317, + "grad_norm": 0.09848370403051376, + "kept_token_count": 573.084375, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.973722267150879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010264758820994757, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.1725299835205, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4393341064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3020, + "valid_token_count": 818.7578125 + }, + { + "actual_token_mask_ratio": 0.30170239210128785, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.015754738356918097, + "grad_norm": 0.10656476020812988, + "kept_token_count": 571.559375, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9757950127124786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.45509375014808e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.49058380126954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.04929809570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3030, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.29992635846138, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.01892449576407671, + "grad_norm": 0.10263603180646896, + "kept_token_count": 572.996875, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.019, + "mgd_cos_sim": 0.976276057958603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.267182540497742e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.61219520568848, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3699432373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3040, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.300832125544548, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.015970187820494176, + "grad_norm": 0.11368206143379211, + "kept_token_count": 571.884375, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9763414025306701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.241655934602022e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.5440185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9911163330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3050, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.3004564344882965, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.01858071153983474, + "grad_norm": 0.12078690528869629, + "kept_token_count": 572.1625, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9748728990554809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.815298108151183e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.05393753051757, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8460327148438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3060, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.3007179856300354, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.01777492808178067, + "grad_norm": 0.11428175866603851, + "kept_token_count": 572.2046875, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9765649795532226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.154322178801521e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.91956329345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.01044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3070, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2996921420097351, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.01803177883848548, + "grad_norm": 0.11240247637033463, + "kept_token_count": 572.8484375, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9755882740020752, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.535857898299583e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.17981224060058, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9632080078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.3002843052148819, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.018816096987575293, + "grad_norm": 0.17584241926670074, + "kept_token_count": 572.415625, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9769513487815857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.003394588944503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.95932388305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1443267822266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3090, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.29933409988880155, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.020089365448802708, + "grad_norm": 0.10676891356706619, + "kept_token_count": 573.1609375, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9757789433002472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.461366644245573e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.18216857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2273406982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3100, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.29953857958316804, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.017122716177254915, + "grad_norm": 0.10740409791469574, + "kept_token_count": 573.2328125, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9760113060474396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.370596162625589e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.74955177307129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8747955322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3110, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.3002043068408966, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.017755051143467426, + "grad_norm": 0.10811600834131241, + "kept_token_count": 572.7359375, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9764976441860199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.180626293527893e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.74031486511231, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.74022216796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3120, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3003817439079285, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.020343981217592955, + "grad_norm": 0.1406615823507309, + "kept_token_count": 572.3890625, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9772131264209747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.90114351932425e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.864501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11437072753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3130, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30044202208518983, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.01784785483032465, + "grad_norm": 0.09709285944700241, + "kept_token_count": 572.421875, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9764124572277069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.213907178491354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.47933197021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9362518310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3140, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3008023172616959, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.017692665848881007, + "grad_norm": 0.09200870245695114, + "kept_token_count": 572.2015625, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9764866530895233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.184924492728896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.89350624084473, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1728851318359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3150, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.2998405426740646, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.018751973006874322, + "grad_norm": 0.16044311225414276, + "kept_token_count": 573.0296875, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9748062372207642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.841329665505327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.33847427368164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.7504638671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3160, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3001315087080002, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.01649873089045286, + "grad_norm": 0.10172785818576813, + "kept_token_count": 572.6828125, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9760257124900817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.364976722281426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.33514213562012, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.6880584716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3170, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.3007418066263199, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.01745613934472203, + "grad_norm": 0.10844887793064117, + "kept_token_count": 572.2703125, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9759356498718261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.400158160133287e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.20706558227539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8928527832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3180, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29977133572101594, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.019006142392754555, + "grad_norm": 0.13137494027614594, + "kept_token_count": 572.9140625, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9757818162441254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.460244473302737e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.05966873168946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5009826660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3190, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.3006925642490387, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.018070036172866823, + "grad_norm": 0.08941944688558578, + "kept_token_count": 572.3171875, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9757370173931121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.477746425545774e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.5038558959961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 350.3131042480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3200, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.30066891610622404, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.018013499211519957, + "grad_norm": 0.11202888935804367, + "kept_token_count": 572.225, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.975562447309494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.545935899950564e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.93873901367188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.94521179199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2991116434335709, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.01673317514359951, + "grad_norm": 0.11301591247320175, + "kept_token_count": 573.453125, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9757712543010711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.464374597882852e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.53186225891113, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.53013916015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3220, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3004645764827728, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.014525248762220144, + "grad_norm": 0.11803822964429855, + "kept_token_count": 572.375, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9771958827972412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.907875744625925e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.88151168823242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.48171081542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3230, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3003882497549057, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.017476196214556695, + "grad_norm": 0.12412268668413162, + "kept_token_count": 572.221875, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9760506689548493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.35522715735715e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.1042594909668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.1469329833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3240, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.2998496204614639, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.017327496968209743, + "grad_norm": 0.15323856472969055, + "kept_token_count": 572.9828125, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9758451998233795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.435495012439787e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.63397674560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1177947998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3250, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29933681190013883, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.018643359746783973, + "grad_norm": 0.12501215934753418, + "kept_token_count": 573.3, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9767794668674469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.070541927940212e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.07317314147949, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6395233154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3260, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3002332776784897, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.017579509969800712, + "grad_norm": 0.1160220131278038, + "kept_token_count": 572.8078125, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9778605103492737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.648256916785613e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.50934829711915, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.1493835449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3270, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.2998674362897873, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.01886670058593154, + "grad_norm": 0.11105514317750931, + "kept_token_count": 572.825, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9758998394012451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.414139422005973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.44743957519532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5712524414063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3280, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2998585909605026, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.022747252695262433, + "grad_norm": 0.14089205861091614, + "kept_token_count": 572.9265625, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9750730752944946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.737096770550125e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 65.18129844665528, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2337219238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3290, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.29951349794864657, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.0178753400221467, + "grad_norm": 0.13330666720867157, + "kept_token_count": 573.2828125, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9761748433113098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.306720821768976e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.76612968444825, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.73768920898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3300, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.30009802579879763, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.01778730982914567, + "grad_norm": 0.11481351405382156, + "kept_token_count": 572.68125, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9774181962013244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.821037990855985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.06900787353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.45679931640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3310, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.29898045063018797, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.02032281570136547, + "grad_norm": 0.1663958728313446, + "kept_token_count": 573.6890625, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9762941718101501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.260105216526426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.7222183227539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.6124298095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3320, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.29890601336956024, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.017115516308695077, + "grad_norm": 0.1313573271036148, + "kept_token_count": 573.6390625, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.974814236164093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.838205514824949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.28638534545898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7794952392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3330, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29914245903491976, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.01850212216377258, + "grad_norm": 0.09269993752241135, + "kept_token_count": 573.2703125, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9770467758178711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.966120003606193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.83301086425782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.87061767578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3340, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.3002750247716904, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.017134695686399935, + "grad_norm": 0.10328077524900436, + "kept_token_count": 572.4078125, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9756196141242981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.523607732262463e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.16051788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1979644775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3350, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.2999037176370621, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.01795026520267129, + "grad_norm": 0.12320651859045029, + "kept_token_count": 572.70625, + "learning_rate": 2.968615892181192e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9751659333705902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.700825976324267e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.94081230163575, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.37668762207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3360, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.29932386577129366, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.019100518990308048, + "grad_norm": 0.11551080644130707, + "kept_token_count": 573.3, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9755145013332367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.564665087964385e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.7094669342041, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6871856689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3370, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3002873569726944, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.015766205638647078, + "grad_norm": 0.10975224524736404, + "kept_token_count": 572.646875, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9763357698917389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.243860840797424e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.65542678833008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.40712890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3380, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3008997917175293, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.015558002330362797, + "grad_norm": 0.10935737192630768, + "kept_token_count": 571.95625, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9762298166751862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.285239575547166e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.51590499877929, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.9267150878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3390, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.29913819432258604, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.01991731971502304, + "grad_norm": 0.1257288008928299, + "kept_token_count": 573.24375, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9764822363853455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.186642128042877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.4952621459961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0405578613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3400, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.29901827275753023, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.015905589796602726, + "grad_norm": 0.12238030135631561, + "kept_token_count": 573.4953125, + "learning_rate": 2.966911280977486e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9770423173904419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.967860994744115e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.54602279663087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.34931640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3410, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.29942195415496825, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.01562183266505599, + "grad_norm": 0.1378328502178192, + "kept_token_count": 573.2390625, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9764931380748749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.182391368085518e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.62673873901367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9225555419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3420, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.299189692735672, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.020107196364551782, + "grad_norm": 0.1398182511329651, + "kept_token_count": 573.353125, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9763433039188385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.240912113455124e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.64037704467773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.89557189941405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3430, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.29897684454917905, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.021460237540304662, + "grad_norm": 0.10394737869501114, + "kept_token_count": 573.378125, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9766813218593597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.108878439292311e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.02495498657227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.38170166015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3440, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.29993063509464263, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.018822990730404852, + "grad_norm": 0.11124736815690994, + "kept_token_count": 573.0796875, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.975518274307251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.563197891111485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.44683151245117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.80296630859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3450, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.29993476867675783, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.0201063035055995, + "grad_norm": 0.13155169785022736, + "kept_token_count": 572.5640625, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9775213181972504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.78075705259107e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.75607070922851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.94215393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3460, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.3003972738981247, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.01644297195598483, + "grad_norm": 0.10976611077785492, + "kept_token_count": 572.5859375, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9750431418418884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.748793308972381e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.44445114135742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2713653564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3470, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.3000331252813339, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.019308620877563954, + "grad_norm": 0.14622516930103302, + "kept_token_count": 573.0328125, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9754753112792969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.579981633578427e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.985917282104495, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5430114746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3480, + "valid_token_count": 818.6578125 + }, + { + "actual_token_mask_ratio": 0.299595034122467, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.016587844211608173, + "grad_norm": 0.13013720512390137, + "kept_token_count": 573.1328125, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9757226347923279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.48336128203664e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.7730812072754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.89637145996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3490, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.3004143476486206, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.017374473996460436, + "grad_norm": 0.1379527449607849, + "kept_token_count": 572.51875, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9770417153835297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.968092079157941e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.60326461791992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9991729736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3500, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2995652049779892, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.017033027671277522, + "grad_norm": 0.09832057356834412, + "kept_token_count": 573.0875, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.976476663351059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.188826224999502e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.74116973876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1615875244141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3510, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30085051357746123, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.01982178268954158, + "grad_norm": 0.1338348090648651, + "kept_token_count": 572.14375, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9765233993530273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.170563134830445e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.71685485839843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.82639465332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3520, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29877917766571044, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.019670231081545354, + "grad_norm": 0.1286545693874359, + "kept_token_count": 573.8765625, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.977127593755722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.934552752180026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.94115829467773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5958618164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3012669950723648, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01759629137814045, + "grad_norm": 0.12525025010108948, + "kept_token_count": 572.075, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9769464552402496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.005308820633217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.32005767822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.5422424316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3540, + "valid_token_count": 818.73125 + }, + { + "actual_token_mask_ratio": 0.30038002133369446, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.018001818843185902, + "grad_norm": 0.11889734119176865, + "kept_token_count": 572.4546875, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9761138081550598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.330566099379212e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.61351013183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4408721923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3550, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30081273913383483, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.01870750803500414, + "grad_norm": 0.18337292969226837, + "kept_token_count": 572.0375, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9757543802261353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.470966324442998e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.16120529174805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.31141967773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3560, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30077717900276185, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.018760222475975752, + "grad_norm": 0.14337459206581116, + "kept_token_count": 572.1, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9769046008586884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.021660225698725e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.75840072631836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 352.54510498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30115216672420503, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.019392091128975154, + "grad_norm": 0.11875785142183304, + "kept_token_count": 572.025, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9758550226688385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.431648431927897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.93378524780273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9977142333984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3580, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.29950237572193145, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.01781104365363717, + "grad_norm": 0.11697712540626526, + "kept_token_count": 573.2703125, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9759497344493866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.39465273404494e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.55813064575196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1923309326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3590, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.30048232078552245, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.019678533636033536, + "grad_norm": 0.1493527591228485, + "kept_token_count": 572.425, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9766927242279053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.104424534598366e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.42941360473633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.78978271484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3600, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29945031106472014, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.017181377857923508, + "grad_norm": 0.12382695078849792, + "kept_token_count": 573.059375, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9764534890651703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.19787256862037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.99697418212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9304473876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3610, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3008485704660416, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.015766760054975747, + "grad_norm": 0.1124047040939331, + "kept_token_count": 571.7859375, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.976789402961731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.066657075891271e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.6393196105957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1820404052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3620, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.2995575308799744, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.01677842903882265, + "grad_norm": 0.13982684910297394, + "kept_token_count": 573.0140625, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9765090823173523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.176158710033633e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.22258758544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3929504394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3630, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30063248574733736, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.0179261714220047, + "grad_norm": 0.12057650834321976, + "kept_token_count": 572.4828125, + "learning_rate": 2.958496961429126e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9760618805885315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.35084855882451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.21819686889648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1693145751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3640, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.30020134449005126, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.0198244608938694, + "grad_norm": 0.17940637469291687, + "kept_token_count": 572.5640625, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9756373882293701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.516661521047354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.28465576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.49352111816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3650, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.3001510679721832, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.018727428372949363, + "grad_norm": 0.1639242172241211, + "kept_token_count": 572.671875, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9762987673282624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.258316422346979e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.48011322021485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6114837646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.30072590708732605, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.019207644276320934, + "grad_norm": 0.12912951409816742, + "kept_token_count": 572.215625, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9761597156524658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.312632682849653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.78514099121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3806518554687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3670, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.30001779496669767, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.016697207745164633, + "grad_norm": 0.14233048260211945, + "kept_token_count": 573.0296875, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9760301828384399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.363227654830552e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.42664642333985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.62128295898435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3680, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.3000784754753113, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.017271273024380206, + "grad_norm": 0.10038823634386063, + "kept_token_count": 572.7125, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9765364050865173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.165489245788194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.49899673461914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.26573486328124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3690, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29942719340324403, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.01947719370946288, + "grad_norm": 0.1082189753651619, + "kept_token_count": 573.4984375, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9766222536563873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.13195515749976e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.1361587524414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6421783447266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3700, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.2996268391609192, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.018167546018958092, + "grad_norm": 0.11474976688623428, + "kept_token_count": 573.0296875, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9757878065109253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.457908890908584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.48119812011718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7942718505859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3710, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3006468594074249, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.01972758201882243, + "grad_norm": 0.12061472237110138, + "kept_token_count": 572.346875, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9760466635227203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.356796072097495e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.31962203979492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.38764038085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3720, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.3011487364768982, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.018602020014077424, + "grad_norm": 0.11479873955249786, + "kept_token_count": 571.8375, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9753470420837402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.630086060496978e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.16201171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.8074493408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3730, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2992502093315125, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.017578540463000537, + "grad_norm": 0.105553537607193, + "kept_token_count": 573.4453125, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9762893497943879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.261990126105957e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.32133483886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1106292724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3740, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29935765266418457, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.019535168074071407, + "grad_norm": 0.19357475638389587, + "kept_token_count": 573.0171875, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9762151002883911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.290993984905072e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.40737075805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57298278808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3750, + "valid_token_count": 817.840625 + }, + { + "actual_token_mask_ratio": 0.30057176649570466, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.018495726957917215, + "grad_norm": 0.14849764108657837, + "kept_token_count": 572.31875, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.975343507528305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.6314638358308e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.74537200927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5304321289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3760, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2989803910255432, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.01747433338314295, + "grad_norm": 0.1048712506890297, + "kept_token_count": 573.4484375, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9770925402641296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.948249742388726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.89158782958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9515411376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3770, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.2993423044681549, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.016285493317991496, + "grad_norm": 0.11982368677854538, + "kept_token_count": 573.4234375, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.976223373413086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.287764914915897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.53098487854004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3020812988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3780, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29919254779815674, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.018204489443451165, + "grad_norm": 0.12715788185596466, + "kept_token_count": 573.35, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9764985799789428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.180264678434468e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.71990127563477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6120178222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3790, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.29983083009719846, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.018715406116098166, + "grad_norm": 0.14930227398872375, + "kept_token_count": 572.8484375, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9748881936073304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.809317198232748e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.78107833862305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6743530273437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3800, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30155782103538514, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.016342022363096474, + "grad_norm": 0.11791955679655075, + "kept_token_count": 571.3375, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9749541997909545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.783533678273671e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.03440780639649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.00111389160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3810, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2987655609846115, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.01850359719246626, + "grad_norm": 0.12656709551811218, + "kept_token_count": 573.8, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9759469032287598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.395759989274665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.98044967651367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1452941894531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3820, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3009074479341507, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.018249430786818267, + "grad_norm": 0.09826068580150604, + "kept_token_count": 572.1734375, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9768792808055877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.031544759636745e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.03532257080079, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2124420166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3830, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.30038482546806333, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.017596003971993922, + "grad_norm": 0.10368436574935913, + "kept_token_count": 572.4375, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9758362531661987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.43897721299436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.56428375244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6389434814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3840, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.29954604506492616, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.01784022441133857, + "grad_norm": 0.10724767297506332, + "kept_token_count": 573.2203125, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9760843873023987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.342055491288193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.89155578613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48338928222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3850, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.30065388381481173, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.01825518561527133, + "grad_norm": 0.12591753900051117, + "kept_token_count": 572.0625, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9752164840698242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.681076553533786e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.14344711303711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53541564941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3860, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30051869750022886, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.01971956416964531, + "grad_norm": 0.12971827387809753, + "kept_token_count": 572.3640625, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9756968677043915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.493428515270353e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.48444328308105, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0893493652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3870, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.29973105192184446, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.018242284189909697, + "grad_norm": 0.11038465052843094, + "kept_token_count": 572.959375, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9756040573120117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.529688686598093e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.03306274414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0067108154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3880, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29983929097652434, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.017095604725182057, + "grad_norm": 0.1452406793832779, + "kept_token_count": 572.9234375, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9737705826759339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010245883895549923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.897525024414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.45152587890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3890, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30003821551799775, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.018885834980756043, + "grad_norm": 0.11924851685762405, + "kept_token_count": 572.8890625, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.973195880651474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010470380293554626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.03200645446778, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.14698181152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3900, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3003260612487793, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.02062184028327465, + "grad_norm": 0.11142360419034958, + "kept_token_count": 572.6171875, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9753397822380065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.632917790440843e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.68365364074707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.25356140136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3910, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.30035060048103335, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.019821873307228087, + "grad_norm": 0.156892791390419, + "kept_token_count": 572.4953125, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9747160911560059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.876549229375087e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.05114974975587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6488006591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3920, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29921679496765136, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.01656120251864195, + "grad_norm": 0.12150859832763672, + "kept_token_count": 573.071875, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.976903909444809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.02193489309866e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.9362419128418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3455047607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3930, + "valid_token_count": 817.759375 + }, + { + "actual_token_mask_ratio": 0.3002989590167999, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.017076065484434366, + "grad_norm": 0.12930478155612946, + "kept_token_count": 572.4578125, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9750337302684784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.752466066856869e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.32924041748046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1183197021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3007091909646988, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.01753542022779584, + "grad_norm": 0.1262294203042984, + "kept_token_count": 572.165625, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9759874880313874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.379904586239717e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.40091400146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1538970947266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3950, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.30026757419109346, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.019418697990477084, + "grad_norm": 0.14182662963867188, + "kept_token_count": 572.48125, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.975535249710083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.556559234624729e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.08133010864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4284393310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3960, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.30000336170196534, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.017647531162947417, + "grad_norm": 0.1349600851535797, + "kept_token_count": 572.7875, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9756883144378662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.49677181779407e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.76129913330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0214904785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2989711731672287, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.02404714245349169, + "grad_norm": 0.14607100188732147, + "kept_token_count": 573.728125, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9758295953273773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.441587390028872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.93565521240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.01568603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3980, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30048043727874757, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.02193604800850153, + "grad_norm": 0.1464962363243103, + "kept_token_count": 572.3296875, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9755490064620972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.551187395118177e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.63621215820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3899871826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3990, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2991031140089035, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.018854919448494913, + "grad_norm": 0.10456699132919312, + "kept_token_count": 573.2453125, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.97697993516922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.992231087177061e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.6577392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8152679443359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4000, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.3008780151605606, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.01970518585294485, + "grad_norm": 0.1359885334968567, + "kept_token_count": 571.971875, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9765180289745331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.172664867946878e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.84661102294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6917236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4010, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2993972599506378, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.01716910917311907, + "grad_norm": 0.14592242240905762, + "kept_token_count": 573.4828125, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9758411467075347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.43707360420376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.2454704284668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9089813232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4020, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.30005186796188354, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.016432957910001277, + "grad_norm": 0.12083785235881805, + "kept_token_count": 572.8984375, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9770471572875976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.965976157924161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.66770935058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.27467956542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4030, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.298904624581337, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.017971811164170504, + "grad_norm": 0.1289401352405548, + "kept_token_count": 573.5046875, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9760234296321869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.365865116706118e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.54682998657226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.98472900390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4040, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.3010235339403152, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.01574100786820054, + "grad_norm": 0.11166631430387497, + "kept_token_count": 572.0515625, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9765584886074066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.156855521723628e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.17339782714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.37890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.2990885555744171, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.017455924302339554, + "grad_norm": 0.1304679661989212, + "kept_token_count": 573.409375, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.976436334848404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.204573798342608e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.75197830200196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8345947265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4060, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.30067485868930816, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.018300541676580906, + "grad_norm": 0.12069854885339737, + "kept_token_count": 572.225, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9757684946060181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.465452239965089e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.83601989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.95209045410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4070, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.2997139245271683, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.017866926919668912, + "grad_norm": 0.10633645206689835, + "kept_token_count": 573.171875, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9763051807880402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.25580992770847e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.20928192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2321411132813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4080, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.29952630698680877, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.0176609817892313, + "grad_norm": 0.1672433465719223, + "kept_token_count": 572.8921875, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9761900126934051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.300793462898582e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.38428192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.500341796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4090, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.3006470263004303, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.018632136285305023, + "grad_norm": 0.11534839123487473, + "kept_token_count": 572.209375, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9767864227294922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.067825085367076e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.14817123413086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0606384277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4100, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.3007981479167938, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.018169183190912007, + "grad_norm": 0.11287394911050797, + "kept_token_count": 572.43125, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9756286442279816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.520084568066522e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.4805908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.66892395019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4110, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.30033630430698394, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.017638429533690214, + "grad_norm": 0.12506994605064392, + "kept_token_count": 572.4515625, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9769877076148987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.989189809653908e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.49323196411133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3308868408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4120, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.29975531101226804, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.01556761534884572, + "grad_norm": 0.11403200030326843, + "kept_token_count": 572.9453125, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.977354782819748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.845804040902294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.40011138916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1120086669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4130, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3001595139503479, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.02156161442399025, + "grad_norm": 0.11031502485275269, + "kept_token_count": 572.7328125, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9763905882835389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.222442167811096e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.01074981689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.166552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4140, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2999875247478485, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.01788499429821968, + "grad_norm": 0.13893333077430725, + "kept_token_count": 572.678125, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9753950238227844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.611337663955055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.55197067260742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3545196533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4150, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.29936380684375763, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.019994108099490406, + "grad_norm": 0.10249673575162888, + "kept_token_count": 573.265625, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9762694299221039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.269770962418988e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.88847198486329, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9403381347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4160, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30004059672355654, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.018799883406609298, + "grad_norm": 0.10152561217546463, + "kept_token_count": 572.56875, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9751197218894958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.718878645799122e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.10378799438476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.160693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4170, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2999467760324478, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.01826581796631217, + "grad_norm": 0.12523502111434937, + "kept_token_count": 572.803125, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9767522513866425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.081171447178349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.12494430541992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 353.806201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4180, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30120340287685393, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.01630791909992695, + "grad_norm": 0.09176580607891083, + "kept_token_count": 572.009375, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9761117219924926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.33137722313404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.72359161376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7159942626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4190, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.30113903284072874, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.016427766997367144, + "grad_norm": 0.12394110858440399, + "kept_token_count": 571.76875, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9763008236885071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.257507044821978e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.98444900512695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0585968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4200, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.29928161203861237, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.016849693097174167, + "grad_norm": 0.10613590478897095, + "kept_token_count": 573.3203125, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9765441179275512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.162476271740161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.2867935180664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.94727783203126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4210, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30064347088336946, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.01627985779196024, + "grad_norm": 0.13093765079975128, + "kept_token_count": 572.4078125, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.976405018568039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.216814214596525e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.01366271972657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.03589782714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4220, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.29974319636821745, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.01726553738117218, + "grad_norm": 0.12089426070451736, + "kept_token_count": 573.2625, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9763576626777649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.235312099917792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.90043411254882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9772491455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4230, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.2990212798118591, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.019482737220823765, + "grad_norm": 0.13134826719760895, + "kept_token_count": 573.6390625, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.978628522157669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.348250048584305e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.63499526977539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.23370056152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4240, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.300291109085083, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.01689969152212143, + "grad_norm": 0.11471067368984222, + "kept_token_count": 572.55625, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9776233077049256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.740915363887325e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.28129348754882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3091186523437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4250, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.29938009679317473, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.016459860280156134, + "grad_norm": 0.11686322838068008, + "kept_token_count": 573.3046875, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9767784953117371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.070924716070294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.15671081542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.44432067871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4260, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.300034499168396, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.014962187875062228, + "grad_norm": 0.10924473404884338, + "kept_token_count": 572.70625, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9761520981788635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.315611896454357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.01871948242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.69927368164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4270, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2990832805633545, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.018312062602490187, + "grad_norm": 0.15700526535511017, + "kept_token_count": 573.58125, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.977596777677536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.751277564442717e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.98046417236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0580780029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4280, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3005359321832657, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.01889452738687396, + "grad_norm": 0.12870241701602936, + "kept_token_count": 572.4203125, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9756446301937103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.51383335632272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.58454971313476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.688037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4290, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.30056828558444976, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.018435663264244794, + "grad_norm": 0.11033829301595688, + "kept_token_count": 572.2265625, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9760736346244812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.346254300908185e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.05736770629883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.52008361816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4300, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2993205636739731, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.01935314079746604, + "grad_norm": 0.12537690997123718, + "kept_token_count": 573.5171875, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.975059962272644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.742218899191357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.92505493164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.30091857910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4310, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.30087166726589204, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.016193137783557177, + "grad_norm": 0.0866512730717659, + "kept_token_count": 572.0, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9756333708763123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.51823138166219e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.50303115844727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.033447265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4320, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2998732358217239, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.018412381783127783, + "grad_norm": 0.12938150763511658, + "kept_token_count": 572.8578125, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9744969189167023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.962164476746693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.39405364990235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.03330993652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4330, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3013285666704178, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.019148453418165446, + "grad_norm": 0.21002493798732758, + "kept_token_count": 571.390625, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9757391214370728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.476931954850442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.3738037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9718780517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4340, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.30016725361347196, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.017675112560391427, + "grad_norm": 0.09747781604528427, + "kept_token_count": 572.6, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9754892468452454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.57452750299126e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.84361572265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6485870361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2986159771680832, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.017475499864667655, + "grad_norm": 0.12324528396129608, + "kept_token_count": 573.771875, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9753721237182618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.62028258072678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.63436508178711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.41655883789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4360, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.3001565903425217, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.018073208164423704, + "grad_norm": 0.16759200394153595, + "kept_token_count": 572.69375, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9770935893058776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.947840397013351e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.01714782714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2657501220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4370, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3000291854143143, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.019540135469287633, + "grad_norm": 0.14224058389663696, + "kept_token_count": 572.73125, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9752672016620636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.66127212450374e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.00544586181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3318328857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4380, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29971869885921476, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.01888818768784404, + "grad_norm": 0.12483760714530945, + "kept_token_count": 573.1875, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9758651435375214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.427698969375343e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.70219345092774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9889801025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4390, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.2988283604383469, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.019759916700422764, + "grad_norm": 0.12825942039489746, + "kept_token_count": 573.959375, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9754231452941895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.600352714187466e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.74315948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5564025878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4400, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.29928006827831266, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.02063641492277384, + "grad_norm": 0.16707807779312134, + "kept_token_count": 573.6015625, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9763766825199127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.2278763622744e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.73465042114258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5376373291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.3000995606184006, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.018848594930022954, + "grad_norm": 0.11204397678375244, + "kept_token_count": 572.978125, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9761869072914123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.302012040279806e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.2324333190918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3546630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4420, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.3001690298318863, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.019267551321536302, + "grad_norm": 0.14334167540073395, + "kept_token_count": 572.6453125, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9759066820144653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.411461433046497e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.41281661987304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.38807678222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4430, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2999700725078583, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.01873215213418007, + "grad_norm": 0.12586648762226105, + "kept_token_count": 573.0109375, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9759020924568176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.413263833266683e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.79108352661133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2722106933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4440, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.2997166395187378, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.01767980828881264, + "grad_norm": 0.13311485946178436, + "kept_token_count": 572.9921875, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9758389472961426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.437927365070209e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.81138343811035, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8789489746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.2997316807508469, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.016143804416060448, + "grad_norm": 0.12404541671276093, + "kept_token_count": 572.690625, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9757647514343262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.466910487390123e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.91768493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2650573730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4460, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.29951401948928835, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.016411618515849113, + "grad_norm": 0.10805351287126541, + "kept_token_count": 573.1359375, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9762179434299469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.289885783800856e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.42114562988282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0436309814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4470, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2994395613670349, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.01904357522726059, + "grad_norm": 0.16422709822654724, + "kept_token_count": 573.2390625, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9753365516662598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.634176676627248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.91621627807618, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.12604370117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4480, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29918733835220335, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.018832398112863302, + "grad_norm": 0.15084487199783325, + "kept_token_count": 573.290625, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9756317257881164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.518874066998251e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.0364601135254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.40247802734376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4490, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.3000029891729355, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.019967895839363335, + "grad_norm": 0.15970218181610107, + "kept_token_count": 572.6703125, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9760213732719422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.366664526169188e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.0458366394043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6911315917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4500, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29959180057048795, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.015271855797618628, + "grad_norm": 0.13241972029209137, + "kept_token_count": 573.1921875, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9767883658409119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.067067294381559e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.87789840698242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.92474670410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4510, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2997283607721329, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.016981753427535294, + "grad_norm": 0.11180736124515533, + "kept_token_count": 572.9359375, + "learning_rate": 2.917702678216765e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9758850455284118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.419918496860191e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.76179351806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.465576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4520, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.3009373933076859, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.016663923859596252, + "grad_norm": 0.11572626233100891, + "kept_token_count": 571.8484375, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9766104638576507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.136552835116163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.01992645263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2449645996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.2999787777662277, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.01812053667381406, + "grad_norm": 0.12355627119541168, + "kept_token_count": 572.7640625, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9755443871021271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.552988849463873e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.2503173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5925354003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4540, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30069233775138854, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.01838180935010314, + "grad_norm": 0.1359613686800003, + "kept_token_count": 572.234375, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9759150803089142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.408184778294525e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.32679443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2772491455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4550, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.29996086955070494, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.018717667646706106, + "grad_norm": 0.12147717922925949, + "kept_token_count": 572.73125, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9780315339565278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.581450310884975e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.38953323364258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0497680664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4560, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2997430056333542, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.016325533948838712, + "grad_norm": 0.12313640862703323, + "kept_token_count": 572.9828125, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9767837584018707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.06885448785033e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.87750091552735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6826110839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4570, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.29776924550533296, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.020004556979984045, + "grad_norm": 0.11165957897901535, + "kept_token_count": 574.3828125, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9770848155021667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.951270356192254e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.51981506347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5311706542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4580, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.2996219038963318, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.020648509077727795, + "grad_norm": 0.13553763926029205, + "kept_token_count": 573.184375, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9757232785224914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.483112080488354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.26148910522461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.37627563476565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4590, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2994832366704941, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.016258358675986528, + "grad_norm": 0.13767817616462708, + "kept_token_count": 573.0203125, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9776525914669036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.72948337928392e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.9799919128418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5029022216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4600, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.2994826972484589, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.02034996822476387, + "grad_norm": 0.13563172519207, + "kept_token_count": 573.190625, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9767121553421021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.096835419768468e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.69705276489258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2104888916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4610, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.30045489966869354, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.01788690537214279, + "grad_norm": 0.12394561618566513, + "kept_token_count": 572.2765625, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.97667316198349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.112065308727323e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.54271392822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2466339111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4620, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.3006542086601257, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.017937665991485118, + "grad_norm": 0.12614105641841888, + "kept_token_count": 572.3359375, + "learning_rate": 2.911656900483265e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9766705811023713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.113074556807988e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.10613403320312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.71013793945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4630, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2999148488044739, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.016251592244952918, + "grad_norm": 0.10115250200033188, + "kept_token_count": 572.9171875, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.977588552236557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.754492664593272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.6130157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5690124511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4640, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30017614364624023, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.017545886989682912, + "grad_norm": 0.10622305423021317, + "kept_token_count": 572.5640625, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9759435892105103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.39705380005762e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.50846481323242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9000701904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2994462639093399, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.016932800970971586, + "grad_norm": 0.12916865944862366, + "kept_token_count": 573.403125, + "learning_rate": 2.909971993098145e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9771330893039704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.93240954610519e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.6074234008789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5474884033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4660, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2999796152114868, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.018512024264782666, + "grad_norm": 0.12998461723327637, + "kept_token_count": 572.7546875, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9776780128479003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.719536126591265e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.87243118286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8717926025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4670, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30032452642917634, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.017850785236805677, + "grad_norm": 0.14109036326408386, + "kept_token_count": 572.646875, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9762966275215149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.259153812308795e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.50406875610352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0262390136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4680, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.3003810852766037, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.019372145738452674, + "grad_norm": 0.14951275289058685, + "kept_token_count": 572.5078125, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9764653861522674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.193229634547606e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.02273406982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3710693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4690, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3002078324556351, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.02007853742688894, + "grad_norm": 0.11653868854045868, + "kept_token_count": 572.5015625, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9766102969646454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.136624576058238e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.22997360229492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.58489379882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4700, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.3006947457790375, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.018705625273287296, + "grad_norm": 0.15136153995990753, + "kept_token_count": 572.2484375, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9758523225784301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.432708247913979e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.32313766479493, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.347216796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4710, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.29964605867862704, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.019192081410437822, + "grad_norm": 0.14739640057086945, + "kept_token_count": 573.409375, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9751720607280732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.698431604192592e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.4104377746582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.18553161621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4720, + "valid_token_count": 818.75 + }, + { + "actual_token_mask_ratio": 0.29972783029079436, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.018646482843905687, + "grad_norm": 0.13601233065128326, + "kept_token_count": 572.8453125, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9768442392349244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.0452324366197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.49397506713868, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2043762207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4730, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.2997718572616577, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.018246709275990723, + "grad_norm": 0.0806363895535469, + "kept_token_count": 572.946875, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9763018488883972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.257104829885066e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.45509567260743, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.44618225097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4740, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2995843827724457, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.02020032927393913, + "grad_norm": 0.14343710243701935, + "kept_token_count": 573.284375, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9751971185207366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.688650316093117e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.85682067871093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.80836791992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4750, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.30099170804023745, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.018396413326263426, + "grad_norm": 0.12371738255023956, + "kept_token_count": 571.8546875, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9762121200561523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.292156173614785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.4700439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.07091369628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.2989684075117111, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.01907151732593775, + "grad_norm": 0.11643014848232269, + "kept_token_count": 573.5046875, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9751141190528869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.721063834149391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.32571029663086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9166534423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4770, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29999094605445864, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.02070405939593911, + "grad_norm": 0.12824246287345886, + "kept_token_count": 572.85, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9759868860244751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.380137489642948e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.21927871704102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.95102844238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4780, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3012510150671005, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.019522691518068312, + "grad_norm": 0.11396599560976028, + "kept_token_count": 571.834375, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9762268126010895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.2864285397809e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.92612228393554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4451904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4790, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30020634531974794, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.016186525113880634, + "grad_norm": 0.1233353540301323, + "kept_token_count": 572.8640625, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9765092372894287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.176097519230098e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.9848258972168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4656066894531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4800, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.2994160592556, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.020633127167820932, + "grad_norm": 0.14732836186885834, + "kept_token_count": 573.36875, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9763669312000275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.231687217834406e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.29573287963868, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3463531494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4810, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.3006753921508789, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.018757030367851257, + "grad_norm": 0.1407240927219391, + "kept_token_count": 572.2796875, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9763101458549499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.253875468857586e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.2682014465332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2517456054687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4820, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.30028662383556365, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.019088692869991063, + "grad_norm": 0.11978083848953247, + "kept_token_count": 572.6921875, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9760934293270112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.33852803427726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.4373291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.15966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4830, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.30002698600292205, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.019541579578071833, + "grad_norm": 0.10623067617416382, + "kept_token_count": 572.790625, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9769074201583863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.020561556098983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.89490051269532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.44891052246095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4840, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2996128350496292, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.01734979944303632, + "grad_norm": 0.110194630920887, + "kept_token_count": 573.1984375, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9768506169319153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.042742676683702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.75857772827149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.36998596191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4850, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.30022495090961454, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.019653524924069642, + "grad_norm": 0.11610785126686096, + "kept_token_count": 572.69375, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9769191741943359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.015969044412486e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.2118766784668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.349072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4860, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.30073895752429963, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.018816589657217265, + "grad_norm": 0.09059961885213852, + "kept_token_count": 572.2046875, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9766614675521851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.116638830164447e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.45773391723633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4963134765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4870, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30029991567134856, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.01803913488984108, + "grad_norm": 0.09857673943042755, + "kept_token_count": 572.434375, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9762896180152894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.261886734748259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.00780334472657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.82012634277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4880, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.3003033518791199, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.017134576011449097, + "grad_norm": 0.0913967564702034, + "kept_token_count": 572.4609375, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.976800388097763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.062368044396863e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.04117965698242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0553802490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4890, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29908743500709534, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.01724861850962043, + "grad_norm": 0.09273094683885574, + "kept_token_count": 573.628125, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9762875139713287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.262704843422398e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.51587677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2966735839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4900, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30046719312667847, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.01687302580103278, + "grad_norm": 0.11259534955024719, + "kept_token_count": 572.2328125, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9765724062919616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.151420526904985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.18592071533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4691925048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4910, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.3001776337623596, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.021771201118826865, + "grad_norm": 0.11136642098426819, + "kept_token_count": 572.44375, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9759382784366608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.399131886311807e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.37759552001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1953826904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4920, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.3005896806716919, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.019677597749978303, + "grad_norm": 0.13010302186012268, + "kept_token_count": 572.3578125, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9787518441677093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.300074987346306e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.58719406127929, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2901580810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4930, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.3008946657180786, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.018564751464873554, + "grad_norm": 0.15333136916160583, + "kept_token_count": 572.184375, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9763594627380371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.234599201590754e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.7101837158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.579150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4940, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.30006212890148165, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.01814937833696604, + "grad_norm": 0.1046491265296936, + "kept_token_count": 572.7140625, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9770337820053101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.971196511993184e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.44025039672852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9505676269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29969179034233095, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.017333142459392548, + "grad_norm": 0.11628952622413635, + "kept_token_count": 573.2015625, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9770845770835876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.951355703175068e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.38416061401367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1070129394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4960, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.29943373799324036, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.01803808696568012, + "grad_norm": 0.1077369973063469, + "kept_token_count": 573.2375, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9762205600738525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.288862202083692e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.59517440795898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.14714965820315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4970, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2996140718460083, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.018454838264733552, + "grad_norm": 0.12827154994010925, + "kept_token_count": 573.028125, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9769875466823578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.98926198715344e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.34102325439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.687890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30079189538955686, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.018603439349681138, + "grad_norm": 0.12546904385089874, + "kept_token_count": 572.075, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9771464884281158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.927175440476276e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.88950653076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.43047790527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4990, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.30023607015609743, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.018010308127850296, + "grad_norm": 0.09703310579061508, + "kept_token_count": 572.796875, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9775175809860229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.782210425124503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.75396575927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9101837158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5000, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.3002867609262466, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.019023940525949003, + "grad_norm": 0.12426707148551941, + "kept_token_count": 572.665625, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9767743825912476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.072523753275163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.22479629516602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.66492919921876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5010, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3006465435028076, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.01972300447523594, + "grad_norm": 0.17173220217227936, + "kept_token_count": 572.2453125, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9764101505279541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.214804886141792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.36047668457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.22446899414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.3008287936449051, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.019062783289700746, + "grad_norm": 0.15295802056789398, + "kept_token_count": 572.321875, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9766044795513154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.138895838987083e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.97160873413085, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.56606140136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5030, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30101770758628843, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.017762553226202727, + "grad_norm": 0.12106001377105713, + "kept_token_count": 572.0875, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9763243317604064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.248323694919237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.77191162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5735565185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5040, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.3003077059984207, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.01728022051975131, + "grad_norm": 0.1346820592880249, + "kept_token_count": 572.775, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9762585401535034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.27402637898922e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.12197723388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.528515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5050, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.2990415632724762, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.016842099279165267, + "grad_norm": 0.14528338611125946, + "kept_token_count": 573.43125, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9767397165298461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.086064528673887e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91721267700196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.16416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5060, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.30011496841907503, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.01661418341100216, + "grad_norm": 0.13855290412902832, + "kept_token_count": 572.8515625, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9772210597991944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.898042651708237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.36489715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.81097106933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5070, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2992299973964691, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.017791545297950507, + "grad_norm": 0.12436901032924652, + "kept_token_count": 573.55625, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9773179113864898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.860209200065584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.70761489868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.64109497070314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.29899208843708036, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.018168372660875322, + "grad_norm": 0.13734687864780426, + "kept_token_count": 573.565625, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9779793620109558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.601833615102805e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.39715881347657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.19019470214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5090, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2993929356336594, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.01792627600952983, + "grad_norm": 0.10543770343065262, + "kept_token_count": 573.1078125, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.018, + "mgd_cos_sim": 0.977422559261322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.819332215352915e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.77954635620117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1461975097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.29968384802341463, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.01939022671431303, + "grad_norm": 0.1396217942237854, + "kept_token_count": 573.0484375, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9778993904590607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.633077013655565e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.62440414428711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.54814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29962779879570006, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.018199920002371073, + "grad_norm": 0.1391458809375763, + "kept_token_count": 573.015625, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9754588067531585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.586423257132993e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.33295669555665, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.74603271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5120, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29930412471294404, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.018007685150951146, + "grad_norm": 0.11909427493810654, + "kept_token_count": 573.4984375, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9771131932735443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.940177940530702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.48684616088867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1501190185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5130, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.300054994225502, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.021541295014321803, + "grad_norm": 0.12822139263153076, + "kept_token_count": 572.7765625, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9743904948234559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00010003727074945346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.2690444946289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.28944702148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5140, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2991742730140686, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.019604308158159257, + "grad_norm": 0.10388638079166412, + "kept_token_count": 573.246875, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9764168500900269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.212182048941031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.58076629638671, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4641479492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5150, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.30006469786167145, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.018812939152121545, + "grad_norm": 0.11495428532361984, + "kept_token_count": 572.9296875, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9765206813812256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.17162309633568e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.16446914672852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.961865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5160, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.30044529139995574, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.017657363228499888, + "grad_norm": 0.12553280591964722, + "kept_token_count": 572.428125, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9778116703033447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.667331567266957e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.91206665039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3653198242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5170, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.2990719318389893, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.01572355842217803, + "grad_norm": 0.11170871555805206, + "kept_token_count": 573.5484375, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.97629314661026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.260512742912397e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.43977661132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8611145019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5180, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2997171193361282, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.019817779306322336, + "grad_norm": 0.1212567612528801, + "kept_token_count": 573.0546875, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9771442294120789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.928058668971062e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.05291442871093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.36207885742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5190, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3000087678432465, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.01729940380901098, + "grad_norm": 0.08768673241138458, + "kept_token_count": 572.8234375, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9766483902931213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.121736584347673e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.04160079956054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78614807128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5200, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.29892840683460237, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.017712619341909885, + "grad_norm": 0.10209094732999802, + "kept_token_count": 573.54375, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9763790607452393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.226951588061638e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.1413345336914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.52073059082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5210, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2991575330495834, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.017573118768632413, + "grad_norm": 0.1166539266705513, + "kept_token_count": 573.49375, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9778304874897004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.659986488055438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.72997283935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2741394042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5220, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30003235638141634, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.01888581896200776, + "grad_norm": 0.1356966197490692, + "kept_token_count": 572.70625, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9767827749252319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.069244624697604e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.77320098876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3912841796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5230, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.30127098858356477, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.01737405201420188, + "grad_norm": 0.12284153699874878, + "kept_token_count": 571.746875, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.976927000284195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.012911323225126e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.90171890258789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0321960449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5240, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2998753696680069, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.01818917114287615, + "grad_norm": 0.10540775954723358, + "kept_token_count": 573.153125, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.977798330783844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.672539406688884e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.31961975097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3775207519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5250, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.2986324310302734, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.020385524537414313, + "grad_norm": 0.17139393091201782, + "kept_token_count": 573.9609375, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9760958254337311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.337586234323681e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.6569107055664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4075592041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5260, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2997500091791153, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.018064386490732432, + "grad_norm": 0.12141722440719604, + "kept_token_count": 573.171875, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.977279144525528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.875353742041626e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.10391159057617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.88441162109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5270, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.3005548745393753, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.01828119708225131, + "grad_norm": 0.13968674838542938, + "kept_token_count": 572.128125, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9769826710224152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.991159047582187e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.65536193847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5751190185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5280, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.30085049867630004, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.016491383034735917, + "grad_norm": 0.09908762574195862, + "kept_token_count": 572.0421875, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9765222847461701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.171005876851268e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.34782638549805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7875915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5290, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3006409704685211, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.017878273781389, + "grad_norm": 0.11176066845655441, + "kept_token_count": 572.45625, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9757893502712249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.457301639486105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.9085906982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7805480957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5300, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2990510821342468, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.01728435941040516, + "grad_norm": 0.0894189178943634, + "kept_token_count": 573.4625, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9769083201885224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.020203506224789e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.40576095581055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9448516845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5310, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2987488001585007, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.018668700382113455, + "grad_norm": 0.1145983561873436, + "kept_token_count": 574.0125, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9759078860282898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.411001738044434e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.291552734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3492004394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5320, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.2989017516374588, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.01661752164363861, + "grad_norm": 0.1126251220703125, + "kept_token_count": 573.821875, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9768316268920898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.0501661907183e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.8863426208496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5035034179688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.29954926669597626, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.01720990929752588, + "grad_norm": 0.15319737792015076, + "kept_token_count": 573.1515625, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9767822325229645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.069458901649341e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.6141128540039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2696807861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5340, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2994417488574982, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.016034358367323875, + "grad_norm": 0.10540907829999924, + "kept_token_count": 573.3671875, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9768138587474823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.057103743543848e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.23101272583008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.74281311035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5350, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.300160625576973, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.01708356598392129, + "grad_norm": 0.12315637618303299, + "kept_token_count": 572.5671875, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9755968153476715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.532509648124687e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.67394104003907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5927032470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5360, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.3000792533159256, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.01633158400654793, + "grad_norm": 0.12210941314697266, + "kept_token_count": 572.6359375, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9766347169876098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.12707662791945e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.38363876342774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.96454162597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5370, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.30017090439796446, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.01760230362415314, + "grad_norm": 0.11234696954488754, + "kept_token_count": 572.56875, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9779578387737274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.610233271610923e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.35408020019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.39271240234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3003769636154175, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.01856595017015934, + "grad_norm": 0.10698249936103821, + "kept_token_count": 572.371875, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9764490187168121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.199614505632781e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.01633911132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.93341369628905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.3013247400522232, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.018488474655896425, + "grad_norm": 0.1151798665523529, + "kept_token_count": 571.9671875, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.977805495262146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.669745875522494e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.69185180664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2183410644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5400, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.3000499665737152, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.017413525842130184, + "grad_norm": 0.11097633093595505, + "kept_token_count": 572.7390625, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9774882435798645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.79367747984361e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.0077522277832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7955352783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2996451616287231, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.01861602384597063, + "grad_norm": 0.10040044039487839, + "kept_token_count": 572.9765625, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.977481347322464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.796375623205676e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.49176406860352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1406219482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5420, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29988565742969514, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.01689628027379513, + "grad_norm": 0.11995557695627213, + "kept_token_count": 572.846875, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9776516437530518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.729849359951913e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.20168914794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.64378356933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5430, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2998733580112457, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.01600735345855355, + "grad_norm": 0.13047417998313904, + "kept_token_count": 572.88125, + "learning_rate": 2.860809579675011e-05, + "loss": 0.016, + "mgd_cos_sim": 0.979467099905014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.020679451874458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.6711669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4088623046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5440, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30150870978832245, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.016015165112912656, + "grad_norm": 0.17398099601268768, + "kept_token_count": 571.5203125, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9771986424922943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.906797811505384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.99921646118165, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5817108154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5450, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29993021190166474, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.015887603722512723, + "grad_norm": 0.11015627533197403, + "kept_token_count": 572.9234375, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9768625617027282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.038081479957327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.40514297485352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4594665527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5460, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2998546749353409, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.017867992538958787, + "grad_norm": 0.1002429947257042, + "kept_token_count": 572.8296875, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9777026534080505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.709910180186853e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.35249786376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9399719238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5470, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.29987322688102724, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.0173173900693655, + "grad_norm": 0.12286904454231262, + "kept_token_count": 572.9078125, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9775969445705414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.751212953939103e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.46397018432617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0878173828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2987110763788223, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.01789053566753864, + "grad_norm": 0.13313953578472137, + "kept_token_count": 573.9375, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9775285959243775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.777913026278839e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.90024337768554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.99547119140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5490, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3007714182138443, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.020597283728420734, + "grad_norm": 0.0934232622385025, + "kept_token_count": 572.290625, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9765500664710999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.160152985714376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.85491104125977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.89210205078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3004773586988449, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.016552893538028003, + "grad_norm": 0.1437612622976303, + "kept_token_count": 572.2953125, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9766023337841034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.139727480942384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.34832000732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9064117431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5510, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.3007769465446472, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.017298348248004913, + "grad_norm": 0.0897151306271553, + "kept_token_count": 572.0703125, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.97775017619133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.691350958542899e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.36247482299805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.91075744628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5520, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.2998971790075302, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.015638001170009373, + "grad_norm": 0.14725086092948914, + "kept_token_count": 572.821875, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9773754477500916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.837738860165701e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.07663421630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3680389404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5530, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3004112482070923, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.019583132956176995, + "grad_norm": 0.1741119623184204, + "kept_token_count": 572.6203125, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9772403597831726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.89050483237952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.15219345092774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.25476989746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5540, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.3001143753528595, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.01812768094241619, + "grad_norm": 0.12882405519485474, + "kept_token_count": 572.5609375, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9772185802459716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.899009990273043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.33038024902343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.40435180664065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5550, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.30027262270450594, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.020637448877096176, + "grad_norm": 0.10264784842729568, + "kept_token_count": 572.8328125, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9769330739974975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.010537614813074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.91528701782227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.5455627441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5560, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.30115391612052916, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01820942563936114, + "grad_norm": 0.11280020326375961, + "kept_token_count": 572.0921875, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9786534905433655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.338497427757829e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.29020233154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3799987792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5570, + "valid_token_count": 818.6265625 + }, + { + "actual_token_mask_ratio": 0.3001635253429413, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.02085979925468564, + "grad_norm": 0.1122804805636406, + "kept_token_count": 572.9140625, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9779705703258514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.605262555647642e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.71362991333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5829803466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5580, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.2988693743944168, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.018142347130924462, + "grad_norm": 0.09593238681554794, + "kept_token_count": 573.64375, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9774118363857269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.823521129670553e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.94226989746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.49416198730466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.300324347615242, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.021183877345174552, + "grad_norm": 0.11605551093816757, + "kept_token_count": 572.7015625, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9768915891647338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.026743355207145e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.35363845825195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5564758300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5600, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.3001594543457031, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.019954053778201343, + "grad_norm": 0.11114423722028732, + "kept_token_count": 572.79375, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9756594955921173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.508031216682866e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.77892150878907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7222198486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2998056560754776, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.017810154519975186, + "grad_norm": 0.09164531528949738, + "kept_token_count": 572.8078125, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9774397134780883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.812632586341351e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.0568634033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.50840148925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5620, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.300619637966156, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.021731926780194043, + "grad_norm": 0.11583501100540161, + "kept_token_count": 572.3171875, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9765984356403351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.141254704445601e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.31835327148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10999450683596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5630, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.299165216088295, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.017003743536770345, + "grad_norm": 0.12492360919713974, + "kept_token_count": 573.4421875, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9759583294391632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.391290368512272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.16249542236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4274047851562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5640, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30012203454971315, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.019433215074241162, + "grad_norm": 0.1097407341003418, + "kept_token_count": 572.8859375, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9756821513175964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.499174630036577e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.53984451293945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1819793701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5650, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.3000014632940292, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.01755131930112839, + "grad_norm": 0.12158413976430893, + "kept_token_count": 572.646875, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9763407230377197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.241928273695521e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.30369644165039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.53965759277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5660, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2996125191450119, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.020159292314201594, + "grad_norm": 0.13477753102779388, + "kept_token_count": 572.959375, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9774676978588104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.801699877949431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.89482345581055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3724700927734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5670, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.30058397352695465, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.0162075100466609, + "grad_norm": 0.11726444214582443, + "kept_token_count": 572.3, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9770671963691712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.958150938269682e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.47027740478515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.60341186523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5680, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2989567518234253, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.019778358936309814, + "grad_norm": 0.12652325630187988, + "kept_token_count": 573.5359375, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9771926581859589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.909141324693337e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.33171691894532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.49090576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30048471689224243, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.015136781707406044, + "grad_norm": 0.11602741479873657, + "kept_token_count": 572.459375, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9784572780132293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.415145857725293e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.74925155639649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0877319335938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5700, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3012326925992966, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.020325643476098776, + "grad_norm": 0.10515470802783966, + "kept_token_count": 571.9921875, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.976991593837738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.987683468149043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.28600616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2018646240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5710, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.29997257590293885, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.018778921384364367, + "grad_norm": 0.1186959519982338, + "kept_token_count": 572.8453125, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9773880004882812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.832823514239862e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.83179626464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1232879638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30031619369983675, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.02259188489988446, + "grad_norm": 0.12320045381784439, + "kept_token_count": 572.3828125, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9770148515701294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.978593614301644e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.8790771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7557891845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5730, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.30105907618999483, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.01885653305798769, + "grad_norm": 0.134552001953125, + "kept_token_count": 571.809375, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9764255225658417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.208800838678144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.66063613891602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.99339599609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.3009893625974655, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.017075159586966038, + "grad_norm": 0.10753244906663895, + "kept_token_count": 572.0921875, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9776665329933166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.724028084543534e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.31116027832032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.95892333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5750, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3007175445556641, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.0165929420851171, + "grad_norm": 0.12108378857374191, + "kept_token_count": 572.0078125, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9783163785934448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.470185493933969e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.05934143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2919555664063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5760, + "valid_token_count": 817.9921875 + }, + { + "actual_token_mask_ratio": 0.2990126132965088, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.017478151246905327, + "grad_norm": 0.08111905306577682, + "kept_token_count": 573.4828125, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9785973429679871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.360432693734765e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.35592880249024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3409851074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5770, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2996205657720566, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.019257314782589675, + "grad_norm": 0.13049398362636566, + "kept_token_count": 573.0375, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9785341382026672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.385117907891982e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.57237243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4690887451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5780, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.30102831721305845, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.01537944208830595, + "grad_norm": 0.08328241854906082, + "kept_token_count": 572.14375, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9784886598587036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.402892854064703e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.37734832763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9658233642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5790, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.3001087725162506, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.01625287104398012, + "grad_norm": 0.08937714993953705, + "kept_token_count": 572.5859375, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9784428119659424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.420785743510351e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.33533172607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0563171386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5800, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.29961870312690736, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.016692334413528444, + "grad_norm": 0.12479566782712936, + "kept_token_count": 572.99375, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9776186287403107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.742743448237888e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.06364364624024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.24974670410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5810, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30014089643955233, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.015889592561870815, + "grad_norm": 0.08909332752227783, + "kept_token_count": 572.5984375, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9775709688663483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.761358985793777e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.82143859863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0704864501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5820, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.30181980729103086, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.017598515003919603, + "grad_norm": 0.10825250297784805, + "kept_token_count": 571.171875, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.977902615070343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.631810051156208e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.9852783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.21571655273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5830, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2999316453933716, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.01656591147184372, + "grad_norm": 0.1463630050420761, + "kept_token_count": 572.85, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.97891646027565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.23578200652264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.47621459960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8030151367187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5840, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30039018094539643, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.017298798356205226, + "grad_norm": 0.10849868506193161, + "kept_token_count": 572.3625, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9765095412731171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.175981613225303e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.24385299682618, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9956298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5850, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.30003631114959717, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.016927690897136926, + "grad_norm": 0.10534324496984482, + "kept_token_count": 572.53125, + "learning_rate": 2.830147042037172e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9782677114009857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.489196115988307e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.81941604614258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.19923095703126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5860, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.2998082101345062, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.016622095927596094, + "grad_norm": 0.11716189980506897, + "kept_token_count": 572.8515625, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9771889925003052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.910570904845372e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.61249084472657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8864471435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5870, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.29944828152656555, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.017749383673071863, + "grad_norm": 0.09978616237640381, + "kept_token_count": 573.13125, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.976523345708847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.170594494207762e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.5300392150879, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4479248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5880, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.30079701244831086, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.015179436281323433, + "grad_norm": 0.10909286141395569, + "kept_token_count": 572.296875, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9775095105171203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.785359968896956e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.71529693603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8940734863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5890, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.29882630705833435, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.016770496033132076, + "grad_norm": 0.08438966423273087, + "kept_token_count": 573.7703125, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9776675045490265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.723653372726403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.19295349121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6955169677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5900, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2994475126266479, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.016948575526475905, + "grad_norm": 0.12281089276075363, + "kept_token_count": 573.0515625, + "learning_rate": 2.826305514790654e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9779265522956848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.622464520158246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.70619049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.40078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.2986413836479187, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.017617708630859853, + "grad_norm": 0.13198542594909668, + "kept_token_count": 573.96875, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.976158183813095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.313236005255021e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.90544052124024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3965270996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5920, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.2994120389223099, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.020031194761395456, + "grad_norm": 0.15764781832695007, + "kept_token_count": 573.3703125, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9765005052089691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.179509797831997e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.84168090820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2388641357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5930, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.3004165828227997, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.01854065591469407, + "grad_norm": 0.13694854080677032, + "kept_token_count": 572.428125, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9766687333583832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.113791456911713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.54222946166992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2037658691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5940, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.3007275551557541, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.020163413789123296, + "grad_norm": 0.14320190250873566, + "kept_token_count": 572.271875, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9772420823574066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.889822129276581e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.74882888793945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8938781738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5950, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.2997151345014572, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.016914618480950595, + "grad_norm": 0.15237490832805634, + "kept_token_count": 573.2265625, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.017, + "mgd_cos_sim": 0.978335040807724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.462896148557775e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.15807266235352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5160430908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5960, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.29926627278327944, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.01722479993477464, + "grad_norm": 0.14145736396312714, + "kept_token_count": 573.378125, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9784956693649292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.400144579354674e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93513641357421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2643646240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5970, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29951147437095643, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.01591828288510442, + "grad_norm": 0.11720586568117142, + "kept_token_count": 573.3796875, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9786695957183837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.332211582455784e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.3405876159668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.26328430175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5980, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.29933987855911254, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.017825362924486398, + "grad_norm": 0.11032165586948395, + "kept_token_count": 573.4671875, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9789979159832001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.203961115214043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.62308197021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1993347167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5990, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2997803002595901, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.01657575834542513, + "grad_norm": 0.14783881604671478, + "kept_token_count": 573.2078125, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9768350958824158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.048814754351043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.00525131225587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.20827026367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6000, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.2993167668581009, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.01885028285905719, + "grad_norm": 0.20665325224399567, + "kept_token_count": 573.1421875, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9773395597934723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.851752136251889e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.07760391235351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8946929931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6010, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.29978619515895844, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.017726715840399265, + "grad_norm": 0.135217547416687, + "kept_token_count": 572.9625, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.977846485376358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.653742697788403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.54261932373046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1866882324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6020, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29991759061813356, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.02005643956363201, + "grad_norm": 0.14343075454235077, + "kept_token_count": 572.7640625, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9770235300064087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.975203527370468e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.31090774536133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4160461425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6030, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.2996554285287857, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.01573915407061577, + "grad_norm": 0.1351833939552307, + "kept_token_count": 572.7953125, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9776367008686065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.735685987630859e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.21550140380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5324462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6040, + "valid_token_count": 817.86875 + }, + { + "actual_token_mask_ratio": 0.3004126787185669, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01813327232375741, + "grad_norm": 0.10304124653339386, + "kept_token_count": 572.5421875, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9778676509857178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.64546891534701e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.80843734741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7421905517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6050, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2992179185152054, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.017083875276148318, + "grad_norm": 0.1106308177113533, + "kept_token_count": 573.5296875, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9780326128005982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.581028596381657e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.5635986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.60859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6060, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2996954172849655, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.01697356691583991, + "grad_norm": 0.09217412769794464, + "kept_token_count": 573.2390625, + "learning_rate": 2.813742464676556e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9765250980854034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.169907352770678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.57680816650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4684112548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6070, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.3001905918121338, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.021050249692052603, + "grad_norm": 0.11779757589101791, + "kept_token_count": 572.4109375, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9771459639072418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.927379240049049e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.36744537353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7072998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6080, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.2995231866836548, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.01809380277991295, + "grad_norm": 0.13686174154281616, + "kept_token_count": 573.0546875, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9777111828327179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.706577864359133e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.3367805480957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.48646850585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6090, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2992581188678741, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.017678744811564683, + "grad_norm": 0.121466264128685, + "kept_token_count": 573.534375, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9786821961402893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.327288160216994e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.77450408935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.66965942382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6100, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2987915962934494, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.02014093603938818, + "grad_norm": 0.12064415961503983, + "kept_token_count": 574.2375, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9771723330020905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.917076120269485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.46667404174805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.16588439941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6110, + "valid_token_count": 818.9234375 + }, + { + "actual_token_mask_ratio": 0.29875433146953584, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.01880712667480111, + "grad_norm": 0.13110877573490143, + "kept_token_count": 573.8546875, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.977992856502533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.596559418947436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.73264236450196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.63804321289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6120, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.29938400983810426, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.01884841648861766, + "grad_norm": 0.12228991091251373, + "kept_token_count": 573.1109375, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9769287407398224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.012231239466928e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.67356185913086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2788421630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6130, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.29976227283477785, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.018607052322477103, + "grad_norm": 0.15837782621383667, + "kept_token_count": 572.8453125, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9787272691726685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.309679906233214e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.58173599243165, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.13600463867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6140, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.30096707344055174, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.015580998081713916, + "grad_norm": 0.10049770772457123, + "kept_token_count": 572.01875, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9773622274398803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.842896349960938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.59674301147462, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9589782714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6150, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30048381388187406, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.017870556004345416, + "grad_norm": 0.16493932902812958, + "kept_token_count": 572.2625, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9768665313720704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.03653257410042e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.79933090209961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2200958251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2992261439561844, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.01602102890610695, + "grad_norm": 0.10543714463710785, + "kept_token_count": 573.4125, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9772795915603638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.875177154550329e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.55308837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3316131591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6170, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3006232500076294, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.0166756140999496, + "grad_norm": 0.12516999244689941, + "kept_token_count": 572.378125, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9763232409954071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.248750648112036e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.68530654907227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6464477539063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6180, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.3008348286151886, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.016822805814445018, + "grad_norm": 0.128800168633461, + "kept_token_count": 572.0171875, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.977470588684082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.800566574791446e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.49140167236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.39781494140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6190, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.3013379335403442, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.017586118075996637, + "grad_norm": 0.1352687031030655, + "kept_token_count": 571.753125, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9772883653640747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.87175410753116e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.94407424926757, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6357666015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6200, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.29994654655456543, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.01755193555727601, + "grad_norm": 0.09952716529369354, + "kept_token_count": 572.7375, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9763666927814484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.231775620719418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.82868576049805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.39288024902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6210, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.29923575520515444, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.01629175301641226, + "grad_norm": 0.10210857540369034, + "kept_token_count": 573.39375, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9767854869365692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.068191502592526e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.93995513916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7743347167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6220, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.3007140845060349, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.01667463071644306, + "grad_norm": 0.12833787500858307, + "kept_token_count": 572.3015625, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9767613410949707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.077618451556191e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.63905868530273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3157623291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.3001448690891266, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.018179642129689456, + "grad_norm": 0.1373067945241928, + "kept_token_count": 572.4609375, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9777130126953125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.705875952728093e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.62896728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2425018310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6240, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.29969177544116976, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.01720209838822484, + "grad_norm": 0.11992048472166061, + "kept_token_count": 573.1328125, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9785854160785675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.365090834558942e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.87361755371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2902374267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6250, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2999165803194046, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.01662577297538519, + "grad_norm": 0.12002302706241608, + "kept_token_count": 572.93125, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9779584109783173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.61001513840165e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.17951965332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.52521057128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6260, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2996803045272827, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.01626420523971319, + "grad_norm": 0.11674775928258896, + "kept_token_count": 572.9484375, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9782474339008331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.497112212353386e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.84966583251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4884948730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6270, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30097604393959043, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.015302407369017602, + "grad_norm": 0.1136094480752945, + "kept_token_count": 571.8125, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9785444736480713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.381079678656533e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.38806533813477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.64085083007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6280, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3012881696224213, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.018626211769878866, + "grad_norm": 0.10953707993030548, + "kept_token_count": 571.9625, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9770848989486695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.951226409408264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.82851028442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.81756286621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6290, + "valid_token_count": 818.596875 + }, + { + "actual_token_mask_ratio": 0.29932667315006256, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.017191433068364858, + "grad_norm": 0.10265428572893143, + "kept_token_count": 573.29375, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9780851304531097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.56050981383305e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.34911193847657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5230407714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6300, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.29966392517089846, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.01628321371972561, + "grad_norm": 0.10476956516504288, + "kept_token_count": 573.1375, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9783058822154999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.474283386021853e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.18874664306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0066680908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6310, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30042214691638947, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.01729656057432294, + "grad_norm": 0.10816621035337448, + "kept_token_count": 572.4515625, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.97739959359169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.828300196910277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.59472351074218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.55596923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6320, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2996592402458191, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.019035227503627537, + "grad_norm": 0.11648737639188766, + "kept_token_count": 572.90625, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9775186359882355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.781802098383196e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.28202514648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.19126892089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6330, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.30051239430904386, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.01657113479450345, + "grad_norm": 0.1097329780459404, + "kept_token_count": 572.5578125, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9778137981891633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.666504509164951e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.01515655517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6340, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.2997923731803894, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.02057274067774415, + "grad_norm": 0.1084437370300293, + "kept_token_count": 572.915625, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9781616687774658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.530619816156104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.3014747619629, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3444458007813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29862593710422514, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.0160080642439425, + "grad_norm": 0.11240892112255096, + "kept_token_count": 573.9234375, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.976747465133667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.083042532438412e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.2036247253418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.18455810546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6360, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30092678368091585, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.016958309337496758, + "grad_norm": 0.1233091652393341, + "kept_token_count": 571.9875, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9780962288379669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.556179527658969e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.42804565429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2311248779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6370, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.29988229274749756, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.017058626189827918, + "grad_norm": 0.11633855849504471, + "kept_token_count": 573.225, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9766230821609497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.131628394243307e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.39601211547851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.69725646972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6380, + "valid_token_count": 818.7484375 + }, + { + "actual_token_mask_ratio": 0.29993757605552673, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.01816457659006119, + "grad_norm": 0.11212403327226639, + "kept_token_count": 572.8125, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9775974750518799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.751002533244901e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.06456832885742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.00501708984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6390, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30015426874160767, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.020408636517822744, + "grad_norm": 0.141904816031456, + "kept_token_count": 572.6109375, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9772570669651032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.883977498044259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.44252853393554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.01312866210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6400, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3003461003303528, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.019263844564557077, + "grad_norm": 0.1732252836227417, + "kept_token_count": 572.496875, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9785314261913299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.386178378714249e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.89429550170898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.17763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30023870170116423, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.01820984445512295, + "grad_norm": 0.10762301832437515, + "kept_token_count": 572.4734375, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.978802752494812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.280192996608093e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.41651992797851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3624298095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6420, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.3000772088766098, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.016458423156291247, + "grad_norm": 0.11426037549972534, + "kept_token_count": 572.6671875, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9776497185230255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.730596091481857e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93220825195313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.215771484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6430, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2996763437986374, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.01866206880658865, + "grad_norm": 0.10822182148694992, + "kept_token_count": 573.0, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9780513346195221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.573721133871004e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.70641632080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4637176513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6440, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3003671824932098, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.018174537736922503, + "grad_norm": 0.0965731143951416, + "kept_token_count": 572.4109375, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.977477365732193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.797922491794452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.87304229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.29968538582324983, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.017984496243298054, + "grad_norm": 0.1750260442495346, + "kept_token_count": 573.009375, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9785578131675721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.375869219889864e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.38847579956055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4244415283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6460, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.3007659941911697, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.016191405523568392, + "grad_norm": 0.13386616110801697, + "kept_token_count": 572.271875, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9777523517608643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.690502145327627e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.70838470458985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6879150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6470, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.3009713888168335, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.016952672321349383, + "grad_norm": 0.1305708885192871, + "kept_token_count": 572.13125, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9773641049861908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.842168390401639e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.83905258178712, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9805633544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6480, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2997476577758789, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.019666590634733437, + "grad_norm": 0.10827581584453583, + "kept_token_count": 573.021875, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9782312572002411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.503436401952058e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.27789077758788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.02077026367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6490, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2998370110988617, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.01831821780651808, + "grad_norm": 0.13279275596141815, + "kept_token_count": 572.771875, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.977620393037796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.742049903958105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.94469985961913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3960296630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6500, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.2997841387987137, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.019767075870186092, + "grad_norm": 0.08917169272899628, + "kept_token_count": 572.9609375, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9784963011741639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.399896032642573e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.77430267333985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0194488525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6510, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2992656290531158, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.017412081081420182, + "grad_norm": 0.0925164446234703, + "kept_token_count": 573.4125, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9776692748069763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.722961938474327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.5539535522461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.65126647949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6520, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30086567997932434, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.01787030613049865, + "grad_norm": 0.12307992577552795, + "kept_token_count": 571.7765625, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9768341183662415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.04919252207037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.43721008300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9904724121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6530, + "valid_token_count": 817.8359375 + }, + { + "actual_token_mask_ratio": 0.29994580149650574, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.015480089746415616, + "grad_norm": 0.11998654901981354, + "kept_token_count": 572.8234375, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9775917172431946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.753253641771152e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 74.41653976440429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.05010986328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6540, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29994092881679535, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.017317799385637045, + "grad_norm": 0.12178432196378708, + "kept_token_count": 572.5609375, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9771623730659484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.920968539314345e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.13433532714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.86749572753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6550, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.300235915184021, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.01764571750536561, + "grad_norm": 0.10897326469421387, + "kept_token_count": 572.6890625, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.97721027135849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.902256231522188e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.84364318847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.70635986328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6560, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29977634847164153, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.01813320303335786, + "grad_norm": 0.09592355042695999, + "kept_token_count": 573.0125, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9775280833244324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.778111368883401e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.44944839477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1729278564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6570, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30007449686527254, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.016228254698216916, + "grad_norm": 0.10084328800439835, + "kept_token_count": 572.7296875, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.977164113521576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.920282707549631e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.94275970458985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2560089111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6580, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2993221551179886, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.01810683701187372, + "grad_norm": 0.09099110215902328, + "kept_token_count": 573.103125, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9785208761692047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.39030799397733e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.02088088989258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.68662719726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6590, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.3009174257516861, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.019112691562622786, + "grad_norm": 0.12897612154483795, + "kept_token_count": 572.121875, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9768093168735504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.058879950316623e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.3377700805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.00925598144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6600, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30141696631908416, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.01838586376979947, + "grad_norm": 0.09429547935724258, + "kept_token_count": 571.534375, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9789895474910736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.207220889744349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.034814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8974914550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.3006103128194809, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.019632071256637573, + "grad_norm": 0.12916336953639984, + "kept_token_count": 572.246875, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9767864048480988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.06783367099706e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.75707092285157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3468353271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6620, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2993089586496353, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.019339924305677415, + "grad_norm": 0.12813404202461243, + "kept_token_count": 573.315625, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9786071360111237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.356605758308433e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.69908905029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.28916625976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6630, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.2994257152080536, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.01926059266552329, + "grad_norm": 0.12613780796527863, + "kept_token_count": 573.515625, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9784181237220764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.43043286295142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.67133941650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8856140136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6640, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.2995551824569702, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.021788716316223145, + "grad_norm": 0.10933127254247665, + "kept_token_count": 573.1609375, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9773351788520813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.853471881593577e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.10824661254883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3861602783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6650, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2996688961982727, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.016296538710594177, + "grad_norm": 0.1308344006538391, + "kept_token_count": 573.0078125, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.977501106262207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.788648119661957e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.91113662719727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8196990966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6660, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3005636423826218, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.019469466619193555, + "grad_norm": 0.1156645193696022, + "kept_token_count": 572.3015625, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9773963093757629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.829581420286559e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.63184509277343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3704833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6670, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2992686003446579, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.014902516081929208, + "grad_norm": 0.10480625182390213, + "kept_token_count": 573.3046875, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9780687868595124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.566902033635415e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.51405487060546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4486602783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6680, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2995673269033432, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.016475728061050177, + "grad_norm": 0.11255593597888947, + "kept_token_count": 573.0265625, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9782427906990051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.498925271851476e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.33232269287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.1959930419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6690, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.30019180178642274, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.017758987098932266, + "grad_norm": 0.11353690177202225, + "kept_token_count": 572.734375, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9776534557342529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.729138353373856e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.58774490356446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4896179199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6700, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30034550130367277, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.019479876291006802, + "grad_norm": 0.10649927705526352, + "kept_token_count": 572.6421875, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9772375881671905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.891588513506576e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.06365127563477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4842956542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6710, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.3007228136062622, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.021028790902346373, + "grad_norm": 0.10695882886648178, + "kept_token_count": 572.453125, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9775028467178345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.787969854893163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.46592407226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.98109436035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.3009447783231735, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.01824086681008339, + "grad_norm": 0.1376766860485077, + "kept_token_count": 572.03125, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9770642518997192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.959293481893837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.33924102783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5248077392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6730, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29977516233921053, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.017424844112247227, + "grad_norm": 0.11952445656061172, + "kept_token_count": 572.753125, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9776347577571869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.736437375773675e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.82044296264648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4900848388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6740, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2989421248435974, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.020820721331983805, + "grad_norm": 0.1536332219839096, + "kept_token_count": 573.8015625, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9770414054393768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.96821598871611e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.18659210205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.25440673828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6750, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.299048525094986, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.019360986724495887, + "grad_norm": 0.16803714632987976, + "kept_token_count": 573.5453125, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9775154531002045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.783039156696759e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.25143127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.71131286621096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6760, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2998627692461014, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.018142250645905734, + "grad_norm": 0.12830914556980133, + "kept_token_count": 572.8921875, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9778397858142853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.656352874822915e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.85451202392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.16395874023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6770, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30135727524757383, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.018704938236624002, + "grad_norm": 0.12441258877515793, + "kept_token_count": 571.51875, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9782098054885864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.511814376106485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.76119842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.19057006835936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6780, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.3004312634468079, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.018323305621743204, + "grad_norm": 0.11149973422288895, + "kept_token_count": 572.446875, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9771384000778198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.930337498895824e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.34292068481446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3762481689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6790, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2998672604560852, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.017691475059837102, + "grad_norm": 0.1340329796075821, + "kept_token_count": 572.8234375, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9764333724975586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.205728129018098e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.60248489379883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.87230529785154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6800, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3002560406923294, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.0197259234264493, + "grad_norm": 0.09634783118963242, + "kept_token_count": 572.8140625, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9765998125076294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.140718684648163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.56777877807617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.28193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6810, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.3003658950328827, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.01998142236843705, + "grad_norm": 0.11918389797210693, + "kept_token_count": 572.3703125, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9755168676376342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.56374205998145e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.55598449707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.11610412597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6820, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30009046792984007, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.016357691399753092, + "grad_norm": 0.10361483693122864, + "kept_token_count": 572.540625, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9782559633255005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.493787463521585e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.00767974853515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.59718933105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6830, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.3005534827709198, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.017974654026329518, + "grad_norm": 0.10534977167844772, + "kept_token_count": 572.4140625, + "learning_rate": 2.74764059131045e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9779048025608063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.630961092421785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.34745178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.47877807617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6840, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.3005062371492386, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.016895593143999577, + "grad_norm": 0.10111592710018158, + "kept_token_count": 572.28125, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9782733142375946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.487005907227285e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.91496124267579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.02524719238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6850, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.29969923496246337, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.01686111753806472, + "grad_norm": 0.10011376440525055, + "kept_token_count": 573.0875, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9777365565299988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.696673830854706e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.58495635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.53742065429685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6860, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.2998922735452652, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.017380553390830757, + "grad_norm": 0.1363738477230072, + "kept_token_count": 572.85625, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9782878994941712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.481315890094266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48830184936523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.62567443847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6870, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2987165510654449, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.018264086917042732, + "grad_norm": 0.14812271296977997, + "kept_token_count": 573.84375, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9785422682762146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.381944935536012e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.69416427612305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5656311035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6880, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29962645173072816, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.016959136817604303, + "grad_norm": 0.09109064191579819, + "kept_token_count": 572.9546875, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9782115161418915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.51114520628471e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.69999847412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4482360839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6890, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2992529958486557, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.018130267411470412, + "grad_norm": 0.11437465250492096, + "kept_token_count": 573.5, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9789699018001556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.214903209591284e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.05614318847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6900, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.30029489994049074, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.019156844168901444, + "grad_norm": 0.12361343950033188, + "kept_token_count": 572.4453125, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9762741923332214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.267913701478392e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.10776443481446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2393859863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6910, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3002765566110611, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.017469536419957875, + "grad_norm": 0.10672880709171295, + "kept_token_count": 572.453125, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9790563464164734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.181135999620892e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.21341323852539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3605194091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6920, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.2998268485069275, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.019011549465358258, + "grad_norm": 0.13386687636375427, + "kept_token_count": 573.04375, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9771674811840058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.918975509004667e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.25405044555664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3064025878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6930, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2993210881948471, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.01700167991220951, + "grad_norm": 0.09957242012023926, + "kept_token_count": 573.478125, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9775659441947937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.763322111917659e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.65940704345704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.23340759277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6940, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.2995805859565735, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.018544137571007015, + "grad_norm": 0.1262732446193695, + "kept_token_count": 573.2359375, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9784475266933441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.418958677793852e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86077117919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.56581115722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6950, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.29950055480003357, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.01827985532581806, + "grad_norm": 0.12148648500442505, + "kept_token_count": 573.16875, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9780186533927917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.586477779317647e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.29579315185546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5308410644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6960, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.3007261365652084, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.018092653434723615, + "grad_norm": 0.09717901051044464, + "kept_token_count": 572.2359375, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9777882516384124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.676487850607373e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.94308471679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.64561157226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3005599409341812, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.018968503549695016, + "grad_norm": 0.09371250867843628, + "kept_token_count": 572.5171875, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9778158247470856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.665720379212871e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.01257934570313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0219482421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6980, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30032148361206057, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.01768639599904418, + "grad_norm": 0.11936818063259125, + "kept_token_count": 572.4796875, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9767558574676514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.079759620362892e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.07957458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.31617431640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6990, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.30029453337192535, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.016694356873631477, + "grad_norm": 0.1006600484251976, + "kept_token_count": 572.7, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9768075823783875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.059564690687694e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.04092330932617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.57737121582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7000, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.3002226144075394, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01894129840657115, + "grad_norm": 0.11753112077713013, + "kept_token_count": 572.4953125, + "learning_rate": 2.731818171982932e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9782616078853608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.49158263008576e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.67321395874023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.8406188964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7010, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.3002372235059738, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.016347005497664213, + "grad_norm": 0.08583429455757141, + "kept_token_count": 572.65625, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9789512515068054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.222187825595029e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.20999145507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0830780029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7020, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.2993008613586426, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.016386141907423733, + "grad_norm": 0.1514069139957428, + "kept_token_count": 573.40625, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9780491113662719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.574584135203622e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5192024230957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8721618652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7030, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.30020020008087156, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.018097985070198774, + "grad_norm": 0.09640603512525558, + "kept_token_count": 572.55625, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9779746234416962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.603680762462318e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.56810455322265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.57904052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7040, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2989465057849884, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.017426679003983737, + "grad_norm": 0.12189442664384842, + "kept_token_count": 573.4546875, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9781272232532501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.544072406948544e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72064971923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9539733886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7050, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.2998164236545563, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.016460780519992113, + "grad_norm": 0.24598795175552368, + "kept_token_count": 572.8453125, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9791068315505982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.161404402926565e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.07199020385742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.63505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7060, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.3007645756006241, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.01947869872674346, + "grad_norm": 0.1182984709739685, + "kept_token_count": 572.259375, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9777553617954254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.689327878528275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.8577896118164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.81316833496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7070, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2998556911945343, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.016385431867092848, + "grad_norm": 0.117306187748909, + "kept_token_count": 572.7078125, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9781098008155823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.55087913805619e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.36596908569337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.09746704101565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30030800998210905, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.017558575421571732, + "grad_norm": 0.14554138481616974, + "kept_token_count": 572.6625, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9788900315761566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.24610426207073e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.33488998413085, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.05869750976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7090, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.29960002899169924, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.016998327802866697, + "grad_norm": 0.11677666008472443, + "kept_token_count": 573.0421875, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.017, + "mgd_cos_sim": 0.977507722377777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.786058315308764e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.41090698242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.80970458984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7100, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.30028883218765257, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.018267886340618135, + "grad_norm": 0.08725416660308838, + "kept_token_count": 572.415625, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.979930716753006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.83958166721277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.56920700073242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.86094665527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7110, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3000081866979599, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.019071249011904, + "grad_norm": 0.126627579331398, + "kept_token_count": 572.6890625, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9774693608283996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.801051008049399e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.80731048583985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9219970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7120, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.30105489790439605, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.01840781969949603, + "grad_norm": 0.13933730125427246, + "kept_token_count": 571.815625, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9783346772193908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.463039048365318e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.83979721069336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.55796508789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7130, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.29869653284549713, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.0170765932649374, + "grad_norm": 0.1550585776567459, + "kept_token_count": 573.775, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9774101197719574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.824187898426317e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.54396286010743, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.24786987304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30030094981193545, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.015636751148849725, + "grad_norm": 0.10211355984210968, + "kept_token_count": 572.4953125, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9781921148300171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.518722170265391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.88369445800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4404357910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7150, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.29966832995414733, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.01824496779590845, + "grad_norm": 0.1422448754310608, + "kept_token_count": 573.1125, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.97736576795578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.84151304489933e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.01698837280273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7749938964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7160, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.30016204714775085, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.019553783908486368, + "grad_norm": 0.1115584596991539, + "kept_token_count": 572.840625, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9771383464336395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.930350886657835e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.01407012939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.27837524414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7170, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2997863799333572, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.018354350607842206, + "grad_norm": 0.11985161900520325, + "kept_token_count": 573.0046875, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9776926040649414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.713842398719863e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83524551391602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3060333251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7180, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29910134673118594, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.018569818791002037, + "grad_norm": 0.09851794689893723, + "kept_token_count": 573.796875, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.976821780204773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.054011243279093e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.86130218505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0405975341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7190, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.2998481124639511, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.016898524668067693, + "grad_norm": 0.15947996079921722, + "kept_token_count": 572.95625, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9775614857673645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.765060192672536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1535041809082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0302764892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7200, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.29983367323875426, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.017468996345996857, + "grad_norm": 0.11685386300086975, + "kept_token_count": 572.846875, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9776357233524322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.736066956771538e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.33106002807617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0965972900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7210, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30084505379199983, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015979814156889916, + "grad_norm": 0.106304831802845, + "kept_token_count": 572.2640625, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9779658377170563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.607114432379603e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.55472869873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.64513244628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7220, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.2999247223138809, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.0184555409476161, + "grad_norm": 0.10505909472703934, + "kept_token_count": 572.734375, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9786442220211029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.342121582245454e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.46670532226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2489715576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7230, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2992357790470123, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.01928769126534462, + "grad_norm": 0.12077534943819046, + "kept_token_count": 573.48125, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9785459399223327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.380511935683899e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.57007293701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.36864013671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7240, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30128195881843567, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.01540300641208887, + "grad_norm": 0.08422312140464783, + "kept_token_count": 571.85, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.979662811756134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.944238386698998e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.87807388305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.57379455566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7250, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2997494041919708, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.01604603286832571, + "grad_norm": 0.10848653316497803, + "kept_token_count": 572.984375, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9789092183113098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.238605078076943e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.81279067993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2525604248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7260, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30008080303668977, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.01878125052899122, + "grad_norm": 0.13765287399291992, + "kept_token_count": 572.596875, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9776215612888336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.741598212509416e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.14998626708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.90643615722655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7270, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.29982411563396455, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.018132376298308373, + "grad_norm": 0.1482340544462204, + "kept_token_count": 573.0046875, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9778283655643463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.660812964080832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.70999145507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.99989013671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7280, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29853902459144593, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.01942194886505604, + "grad_norm": 0.14671671390533447, + "kept_token_count": 573.8953125, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.977510416507721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.785014215391129e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.46570358276367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2643615722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7290, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.300214147567749, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.01703766528517008, + "grad_norm": 0.10599256306886673, + "kept_token_count": 572.571875, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9792591273784638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.101920975605026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.19246139526368, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8727111816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7300, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2994201749563217, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.018292201310396196, + "grad_norm": 0.12350074201822281, + "kept_token_count": 573.403125, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9774402260780335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.812429077806883e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.33870925903321, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2596099853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7310, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.29934931695461275, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.019437053892761468, + "grad_norm": 0.11145143210887909, + "kept_token_count": 573.234375, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9784153580665589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.43152120069135e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.1098243713379, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0464385986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7320, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2999861389398575, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.01656891368329525, + "grad_norm": 0.09369955956935883, + "kept_token_count": 572.534375, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.977192348241806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.909259122447111e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.65739364624024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1314331054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7330, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.30037895739078524, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.017641009390354158, + "grad_norm": 0.104629285633564, + "kept_token_count": 572.621875, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9769231915473938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.014395109261387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.39398574829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.703369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7340, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.30041339099407194, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.016476756148040294, + "grad_norm": 0.09643811732530594, + "kept_token_count": 572.5625, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9776833534240723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.717461605556309e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.73514175415039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.78717651367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30010223388671875, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.021343018859624863, + "grad_norm": 0.10519770532846451, + "kept_token_count": 572.671875, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9769514381885529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.003367522382177e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.1350212097168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7871826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7360, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.3004471570253372, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.02103931028395891, + "grad_norm": 0.13035716116428375, + "kept_token_count": 572.4796875, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9764652192592621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.193287478410639e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.9246337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.43316345214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7370, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.29965412616729736, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.017641477193683384, + "grad_norm": 0.164508655667305, + "kept_token_count": 573.178125, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9773774325847626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.836963170324452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.86002731323242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3199462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7380, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2998152494430542, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.0197706307284534, + "grad_norm": 0.15170879662036896, + "kept_token_count": 573.09375, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9783134222030639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.471342225675471e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.90067138671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.08513793945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.3002436518669128, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.016495543532073498, + "grad_norm": 0.12808910012245178, + "kept_token_count": 572.6390625, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9790323317050934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.190510307031218e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.50721588134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9621307373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7400, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30099889934062957, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.01656839419156313, + "grad_norm": 0.12552385032176971, + "kept_token_count": 571.8640625, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9801963150501252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.735835242783651e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96371612548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.38684997558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7410, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2996888518333435, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.016822175402194262, + "grad_norm": 0.10409965366125107, + "kept_token_count": 573.0890625, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9790438354015351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.18602042272687e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.65503540039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.455859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7420, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29974085092544556, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.017707545962184667, + "grad_norm": 0.08975473791360855, + "kept_token_count": 573.1515625, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9777533113956451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.69013434567023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.73494949340821, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5222900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7430, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.3005801737308502, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.017399189155548812, + "grad_norm": 0.11552007496356964, + "kept_token_count": 572.26875, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9790850460529328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.16992120235227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5294578552246, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.12578735351565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7440, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.29957498610019684, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.017527418397367, + "grad_norm": 0.10357393324375153, + "kept_token_count": 573.196875, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9777388036251068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.69579591380898e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.9230972290039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6483459472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7450, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3004676342010498, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.01901395693421364, + "grad_norm": 0.12151894718408585, + "kept_token_count": 572.1671875, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9783358633518219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.462571568088607e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.8569320678711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6998260498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7460, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.3001465916633606, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.017893829476088287, + "grad_norm": 0.10290607064962387, + "kept_token_count": 572.51875, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9781606435775757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.531015846529044e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.84045791625977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.21412048339846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7470, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3012863308191299, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.01619600486010313, + "grad_norm": 0.11003942042589188, + "kept_token_count": 571.9453125, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9775863230228424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.755363814998418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2812286376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.71532287597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7480, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.30033419132232664, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015590894408524037, + "grad_norm": 0.09056153148412704, + "kept_token_count": 572.503125, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9777461290359497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.692934788996353e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.63483963012695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.46169128417966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7490, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30008457899093627, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.01934587024152279, + "grad_norm": 0.1124168336391449, + "kept_token_count": 572.7234375, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9767531216144562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.080832242034376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.81441650390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.69005126953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7500, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29993808567523955, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.016129259765148164, + "grad_norm": 0.11507917940616608, + "kept_token_count": 572.8296875, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9776397347450256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.73450277140364e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.76433792114258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9624053955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7510, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2999284744262695, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.019123153761029243, + "grad_norm": 0.10476028919219971, + "kept_token_count": 572.8109375, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9774330973625183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.81521547853481e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.90851287841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7509307861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7520, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30057630240917205, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.016338798869401216, + "grad_norm": 0.11270622164011002, + "kept_token_count": 572.1390625, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.978262197971344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.491344924550503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.45212631225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.82451782226565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.29960030913352964, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.0157529448159039, + "grad_norm": 0.0985555648803711, + "kept_token_count": 573.328125, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9787831664085388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.287842356367037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.25080947875976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3063537597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7540, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.30014811754226683, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.016103968117386104, + "grad_norm": 0.12166901677846909, + "kept_token_count": 572.984375, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9789784967899322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.211540116462857e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.28499755859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0162322998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7550, + "valid_token_count": 818.71875 + }, + { + "actual_token_mask_ratio": 0.30139324963092806, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.020075517054647208, + "grad_norm": 0.11563844978809357, + "kept_token_count": 571.6140625, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9781959593296051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.517217574990354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.16338272094727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0520294189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7560, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.30087604820728303, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.015588017832487821, + "grad_norm": 0.11267103999853134, + "kept_token_count": 572.203125, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9785559177398682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.376617333851754e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.74529495239258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7219268798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7570, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.29966704845428466, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.015303158853203058, + "grad_norm": 0.11117573082447052, + "kept_token_count": 572.7046875, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9786803662776947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.327998220920562e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44859008789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2547119140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7580, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.3001809507608414, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.016345477383583783, + "grad_norm": 0.09848557412624359, + "kept_token_count": 572.7375, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9793224811553956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.077176971710287e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.62768630981445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.97254943847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7590, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3000236749649048, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.017977290600538254, + "grad_norm": 0.12231359630823135, + "kept_token_count": 572.7609375, + "learning_rate": 2.673586215864882e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9784845292568207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.404497857554816e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.19224700927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.80267639160155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7600, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29956528842449187, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.01623145109042525, + "grad_norm": 0.12175053358078003, + "kept_token_count": 573.2125, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9777475237846375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.692395713296718e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.94515762329101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.717724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7610, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2994926065206528, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.016500675305724143, + "grad_norm": 0.10167960077524185, + "kept_token_count": 573.2640625, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9787847101688385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.287239543278702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.11924209594727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9104309082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7620, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.2994380384683609, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.01760954624041915, + "grad_norm": 0.11049513518810272, + "kept_token_count": 573.3109375, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9787762224674225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.290555633720942e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.18823852539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1135528564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7630, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.29923240542411805, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.017754970490932463, + "grad_norm": 0.12209444493055344, + "kept_token_count": 573.7171875, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9791634619235993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.139289784594438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.67176818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7770172119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7640, + "valid_token_count": 818.7 + }, + { + "actual_token_mask_ratio": 0.30032039880752565, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.01731544258072972, + "grad_norm": 0.11367032676935196, + "kept_token_count": 572.3984375, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9769687056541443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.996618416858836e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.23060989379883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.816943359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7650, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.3003486156463623, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.021086085867136718, + "grad_norm": 0.13540132343769073, + "kept_token_count": 572.5890625, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9769830048084259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.991033901111223e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.05498962402343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2545593261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7660, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29974873661994933, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.017756599560379983, + "grad_norm": 0.17657899856567383, + "kept_token_count": 573.1015625, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9774769604206085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.79808299941942e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.90586929321289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1952850341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7670, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.30033931136131287, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016799585428088903, + "grad_norm": 0.11581762135028839, + "kept_token_count": 572.5828125, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9785234332084656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.38930209283717e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.56271133422851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.10023498535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7680, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3006883680820465, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.015564247407019139, + "grad_norm": 0.13744883239269257, + "kept_token_count": 572.21875, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9781537234783173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.53371930134017e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.54502944946289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3965576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2998808264732361, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.020988225657492875, + "grad_norm": 0.0911378264427185, + "kept_token_count": 572.775, + "learning_rate": 2.663217516246862e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9773808896541596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.835612607072107e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.19415435791015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.15773315429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7700, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3006184011697769, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.018256898783147336, + "grad_norm": 0.12089141458272934, + "kept_token_count": 572.1578125, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9778879225254059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.637556165922434e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.79459686279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6014770507812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7710, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3001044303178787, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.015908768866211176, + "grad_norm": 0.11427689343690872, + "kept_token_count": 572.7765625, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9771993100643158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.906534349080175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.00659408569337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.27430114746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7720, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3009951293468475, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.018508812971413137, + "grad_norm": 0.12859992682933807, + "kept_token_count": 571.9359375, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9782553136348724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.494036082993262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.28595657348633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3963623046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7730, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29965258836746217, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.01610295260325074, + "grad_norm": 0.13885441422462463, + "kept_token_count": 573.3765625, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9787827551364898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.288007229566574e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.55366973876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7798767089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7740, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.2991327404975891, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.017037645727396012, + "grad_norm": 0.10975725203752518, + "kept_token_count": 573.60625, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.978003865480423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.592262456659227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.13221588134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0769897460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7750, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.30002923011779786, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.017389474902302027, + "grad_norm": 0.1152682825922966, + "kept_token_count": 572.7046875, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9769194781780243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.015844843816013e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.30866012573242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.79790954589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7760, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.29943851828575135, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.02247317964211106, + "grad_norm": 0.13360992074012756, + "kept_token_count": 573.4375, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9776826679706574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.717732998775318e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.29727783203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5936553955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7770, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.29991099834442136, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.016676668729633092, + "grad_norm": 0.09772999584674835, + "kept_token_count": 572.9953125, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9785170435905457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.391798182856291e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.72715835571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.19130554199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2994400173425674, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016227680072188378, + "grad_norm": 0.12186744809150696, + "kept_token_count": 573.134375, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9773979127407074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.828961290419102e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.00915298461913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4342987060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7790, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.299430787563324, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.015401803329586983, + "grad_norm": 0.11767928302288055, + "kept_token_count": 573.1640625, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9788574516773224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.258832822320983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.37210083007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2348358154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7800, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.300016850233078, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.01743876188993454, + "grad_norm": 0.10734420269727707, + "kept_token_count": 572.8734375, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9780953526496887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.5565191693604e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.44551467895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9640350341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7810, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.3000055342912674, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.014689270313829184, + "grad_norm": 0.11448141932487488, + "kept_token_count": 572.8796875, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9793290257453918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.074614306678996e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.22726745605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.98408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7820, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.29926508069038393, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.019309700466692448, + "grad_norm": 0.11694005131721497, + "kept_token_count": 573.453125, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9772499382495881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.886764917406253e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77246551513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4074737548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7830, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.299618735909462, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.016680732648819686, + "grad_norm": 0.11211573332548141, + "kept_token_count": 573.35625, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9790668725967407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.177023410098627e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.08958129882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7241577148437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7840, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.30038722455501554, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.01700913570821285, + "grad_norm": 0.1538744568824768, + "kept_token_count": 572.4484375, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9783947765827179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.439560479018837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.19407348632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8511596679688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7850, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.3004970222711563, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.01773169282823801, + "grad_norm": 0.11583907902240753, + "kept_token_count": 572.3703125, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9789318144321442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.229781087720767e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.82131881713867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9745666503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.29941571354866026, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.016205867286771536, + "grad_norm": 0.13167105615139008, + "kept_token_count": 573.1640625, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9784336924552918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.424357511103154e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.32551574707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1195861816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7870, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.301183620095253, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.017682911828160286, + "grad_norm": 0.11129415035247803, + "kept_token_count": 571.7328125, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9789647996425629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.216891073971055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.07872848510742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.69049072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7880, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.30019381642341614, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.01834183493629098, + "grad_norm": 0.1151316836476326, + "kept_token_count": 572.4921875, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9776716768741608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.722016209503636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.61237487792968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.243701171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7890, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.30049446821212766, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.01829669615253806, + "grad_norm": 0.14990343153476715, + "kept_token_count": 572.26875, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9783174991607666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.469738968415186e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.77690963745117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7591491699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7900, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29964151680469514, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.017825658060610296, + "grad_norm": 0.11427880078554153, + "kept_token_count": 573.084375, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9787613153457642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.296380328829401e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.03399276733398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0203399658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7910, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29974326193332673, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.01665798919275403, + "grad_norm": 0.09639492630958557, + "kept_token_count": 572.9515625, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9794565737247467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.024790586205199e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.89432220458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2210754394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7920, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2992518573999405, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.01749118501320481, + "grad_norm": 0.09674584120512009, + "kept_token_count": 573.453125, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9793606400489807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.062267006607726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.86213684082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.042724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7930, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3004341691732407, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.017463790718466043, + "grad_norm": 0.1293092966079712, + "kept_token_count": 572.259375, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9779396176338195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.617349885753356e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.91632919311523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4277801513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7940, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.29957345128059387, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.018132889922708274, + "grad_norm": 0.1294417381286621, + "kept_token_count": 573.1140625, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9777594149112702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.687750014360062e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.716845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3731384277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30025817155838014, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.019206991698592902, + "grad_norm": 0.10231859982013702, + "kept_token_count": 572.5171875, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9793600559234619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.0624994734535e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.1343605041504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.681884765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7960, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.3011104941368103, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.01999611547216773, + "grad_norm": 0.11441013216972351, + "kept_token_count": 571.8375, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9770147383213044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.9786385797197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23404922485352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.41656494140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30090884268283846, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.017238650377839802, + "grad_norm": 0.11359678208827972, + "kept_token_count": 571.9078125, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9786329984664917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.346510730916634e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.6487060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9358245849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7980, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.300080743432045, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.019370604678988455, + "grad_norm": 0.11101755499839783, + "kept_token_count": 572.753125, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9765123307704926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.174887381959706e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.22484283447265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.66200256347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7990, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.300170361995697, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.017251999862492085, + "grad_norm": 0.09807754307985306, + "kept_token_count": 572.6515625, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9763989746570587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.21917126106564e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.28854522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.80971374511716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8000, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3001988857984543, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.01573209445923567, + "grad_norm": 0.1566239446401596, + "kept_token_count": 572.646875, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9784984052181244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.399079160881229e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.02605819702148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7564636230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8010, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.29879908859729765, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.01646373709663749, + "grad_norm": 0.1156589537858963, + "kept_token_count": 573.84375, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9786779165267945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.328960975632072e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.08753967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.08192749023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8020, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29900469779968264, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.017059803754091264, + "grad_norm": 0.10827162116765976, + "kept_token_count": 573.703125, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9780654847621918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.568185512558557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.99226989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.50929870605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8030, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.30008488297462466, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.01775106256827712, + "grad_norm": 0.1089523509144783, + "kept_token_count": 572.990625, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9778176605701446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.664989654789678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.44741134643554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.118310546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8040, + "valid_token_count": 818.665625 + }, + { + "actual_token_mask_ratio": 0.29944959580898284, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.01802424490451813, + "grad_norm": 0.13843943178653717, + "kept_token_count": 573.321875, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9770872890949249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.950300107244403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36897277832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.32570190429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8050, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30060664415359495, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.017422615736722945, + "grad_norm": 0.12589512765407562, + "kept_token_count": 572.3359375, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9782263159751892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.505365840392188e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.49957580566407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5653442382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8060, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3005858510732651, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.016778994537889957, + "grad_norm": 0.13589400053024292, + "kept_token_count": 572.1890625, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9776173114776612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.743259531911463e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.40960083007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.72784729003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8070, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29991053938865664, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.017986764013767243, + "grad_norm": 0.10913987457752228, + "kept_token_count": 572.9328125, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9785265624523163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.388082424062304e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.26646881103515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.0457733154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8080, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.29901280999183655, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.01673372630029917, + "grad_norm": 0.14106684923171997, + "kept_token_count": 573.5953125, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9785683989524842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.371738658752292e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.17467498779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1197448730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8090, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.29991107285022733, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.019682434294372796, + "grad_norm": 0.12729334831237793, + "kept_token_count": 572.765625, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9765310764312745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.167562384391204e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.83960418701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7215606689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8100, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.30001671314239503, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.01644018795341253, + "grad_norm": 0.1365293264389038, + "kept_token_count": 572.884375, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9771356105804443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.931426418712363e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.70222473144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8110, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.29971724152565005, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.019141466915607454, + "grad_norm": 0.14888493716716766, + "kept_token_count": 572.746875, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9771448314189911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.927821108954959e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.93054885864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.15293273925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8120, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.29960982501506805, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.018745720759034155, + "grad_norm": 0.13770772516727448, + "kept_token_count": 572.9984375, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9780894815921783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.558817353332415e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.48025360107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.23985290527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8130, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3005439192056656, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.015385258197784423, + "grad_norm": 0.0960177481174469, + "kept_token_count": 572.2515625, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9779579222202301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.610197401139885e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.31807861328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2190246582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8140, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.30135454535484313, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.018404834903776646, + "grad_norm": 0.12572546303272247, + "kept_token_count": 571.6484375, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9796242237091064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.959303693496622e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.96992568969726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2229064941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8150, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2994381844997406, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.018338847160339355, + "grad_norm": 0.10189320892095566, + "kept_token_count": 573.3359375, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9775865018367768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.75529192853719e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.17411270141602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4300262451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8160, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.29996815621852874, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.01715990174561739, + "grad_norm": 0.08885955065488815, + "kept_token_count": 573.028125, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9781460583209991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.536717286915519e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.4830451965332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.10772399902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8170, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.29999635815620423, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.015574006456881761, + "grad_norm": 0.1062777042388916, + "kept_token_count": 572.5046875, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9797629177570343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.905137463239952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.30372695922851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3325622558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8180, + "valid_token_count": 817.8578125 + }, + { + "actual_token_mask_ratio": 0.3000611662864685, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.020100762508809565, + "grad_norm": 0.11372079700231552, + "kept_token_count": 572.75625, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9768034100532532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.061186356120742e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.05278549194335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5237640380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8190, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30066837966442106, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.016620496474206447, + "grad_norm": 0.09647174179553986, + "kept_token_count": 572.1875, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9785379528999328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.383632593904622e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.30350036621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.859033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.29966476261615754, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.017177462112158537, + "grad_norm": 0.11947501450777054, + "kept_token_count": 572.95, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9780904471874237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.55843914905563e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.53696670532227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.00234069824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8210, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.30023919641971586, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.016527142189443112, + "grad_norm": 0.12272142618894577, + "kept_token_count": 572.6375, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9793120801448822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.08123433671426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.29396896362304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.44443359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2994582235813141, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.016554952785372733, + "grad_norm": 0.13123653829097748, + "kept_token_count": 573.3828125, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9781343877315521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.541280403733254e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.1830207824707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7122344970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8230, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.30054031014442445, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.016762878093868495, + "grad_norm": 0.09091033041477203, + "kept_token_count": 572.4265625, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.979552435874939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.987346980371512e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55582580566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8829742431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8240, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.3008326470851898, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.01784216882660985, + "grad_norm": 0.1112045869231224, + "kept_token_count": 572.184375, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9794605910778046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.023220580071211e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.73630752563477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4326568603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8250, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.29980129897594454, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.017113908007740975, + "grad_norm": 0.16192269325256348, + "kept_token_count": 572.6359375, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9779677987098694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.60634223499801e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.0022804260254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9236785888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8260, + "valid_token_count": 817.8125 + }, + { + "actual_token_mask_ratio": 0.2996375322341919, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.016317000798881053, + "grad_norm": 0.1006770059466362, + "kept_token_count": 573.1859375, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9788241505622863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.271829574368894e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.37835464477538, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6831512451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8270, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.29959006905555724, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.01831575818359852, + "grad_norm": 0.07908933609724045, + "kept_token_count": 573.2390625, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9790740728378295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.174206304829567e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.30105972290039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7425018310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8280, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2994464308023453, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.018529440741986036, + "grad_norm": 0.16669543087482452, + "kept_token_count": 573.5125, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9776680827140808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.723429637029767e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.20192642211914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.93787536621096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8290, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.30132718980312345, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.017236323188990356, + "grad_norm": 0.12050355225801468, + "kept_token_count": 571.7296875, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9784689128398896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.410599257331341e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.12424545288086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.91399536132815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8300, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2993972718715668, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.019117273576557636, + "grad_norm": 0.12098874151706696, + "kept_token_count": 573.3609375, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9791799664497376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.132847724482418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.11659393310546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.82472839355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8310, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29991271793842317, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.0172801585868001, + "grad_norm": 0.13058826327323914, + "kept_token_count": 572.80625, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9775084912776947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.785765021457337e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.80019607543946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.39918518066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8320, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.300984126329422, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.01861142981797457, + "grad_norm": 0.1292252242565155, + "kept_token_count": 571.7484375, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9786909759044647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.32385485409759e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.27913436889648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.288037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8330, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.3008982419967651, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.016981802042573692, + "grad_norm": 0.11097072064876556, + "kept_token_count": 571.959375, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9781582772731781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.531941057299264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.93732147216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.02549743652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8340, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.299594047665596, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.019638343527913094, + "grad_norm": 0.13265646994113922, + "kept_token_count": 572.9953125, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9774769842624664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.798071503406391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.56224212646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9356323242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8350, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.3004858911037445, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.017155392933636902, + "grad_norm": 0.15556418895721436, + "kept_token_count": 572.4140625, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9789944291114807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.205323611036875e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.0978630065918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2450897216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8360, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.2990195691585541, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.014908172190189362, + "grad_norm": 0.08362343162298203, + "kept_token_count": 573.375, + "learning_rate": 2.590164913622729e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9781453490257264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.536986424587668e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.08279724121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.01629638671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8370, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.30024394094944, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.017356527969241142, + "grad_norm": 0.10622981935739517, + "kept_token_count": 572.5671875, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9787776529788971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.289999386761338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.43498001098632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.192822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8380, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2985456079244614, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.01691081179305911, + "grad_norm": 0.08753664046525955, + "kept_token_count": 573.8703125, + "learning_rate": 2.587890815618397e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9794679522514343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.020350796869024e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25969009399414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.26195373535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8390, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3009283900260925, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.01629997631534934, + "grad_norm": 0.09499607235193253, + "kept_token_count": 571.965625, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9794878721237182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.012567268451676e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.65112609863282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.72719421386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8400, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2997672617435455, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.01855015093460679, + "grad_norm": 0.12847201526165009, + "kept_token_count": 572.98125, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.97911217212677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.159321077982895e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.28273544311523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.59190979003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8410, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30051379799842837, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.017127500101923942, + "grad_norm": 0.13799770176410675, + "kept_token_count": 572.3203125, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9786619961261749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.335176680702716e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.38011093139649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4556365966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8420, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29983149468898773, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.015747414249926806, + "grad_norm": 0.11665641516447067, + "kept_token_count": 573.0578125, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9784402668476104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.421788297710008e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.46599578857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1709747314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8430, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.299272483587265, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.016765307262539865, + "grad_norm": 0.11997620761394501, + "kept_token_count": 573.41875, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9793817520141601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.054021527641453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.78406600952148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3993408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8440, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30104056000709534, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.020408345106989145, + "grad_norm": 0.12175427377223969, + "kept_token_count": 572.13125, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.978606766462326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.356746548088267e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44414443969727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.16880493164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8450, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.30024094581604005, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.01692581893876195, + "grad_norm": 0.11819767206907272, + "kept_token_count": 572.6375, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9779707491397858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.6051961989142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.96489105224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4323455810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8460, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30075023174285886, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.0177931385114789, + "grad_norm": 0.13686232268810272, + "kept_token_count": 571.9265625, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9790902316570282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.167897904058919e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.78932266235351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.03883666992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8470, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.3013390123844147, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.01619268087670207, + "grad_norm": 0.11522461473941803, + "kept_token_count": 571.753125, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9780620753765106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.569522542529739e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.89016723632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.14459228515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8480, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.29945493340492246, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.016978134401142597, + "grad_norm": 0.1342000812292099, + "kept_token_count": 573.0796875, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9795431017875671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.990999511093832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.05358123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.70625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8490, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.30086599886417387, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.018094115890562534, + "grad_norm": 0.1275235265493393, + "kept_token_count": 571.9578125, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9782200694084168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.507803722750396e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.68185882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5737762451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8500, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.300109001994133, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.01667377334088087, + "grad_norm": 0.13122282922267914, + "kept_token_count": 572.6265625, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9785006999969482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.398178324569017e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.92481918334961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.165771484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8510, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30018181502819063, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.016382378805428744, + "grad_norm": 0.10833191871643066, + "kept_token_count": 572.7046875, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.978281807899475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.483689816785045e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.78769607543946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8199737548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8520, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29964573085308077, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.017160367220640182, + "grad_norm": 0.1374167650938034, + "kept_token_count": 573.16875, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9796126127243042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.96384374552872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.52503356933593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.97479553222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8530, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3006525307893753, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.01836447510868311, + "grad_norm": 0.12491663545370102, + "kept_token_count": 572.0484375, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9783263683319092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.466278959531336e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.50838165283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.03063049316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8540, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.2995960354804993, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.015392777603119611, + "grad_norm": 0.10743732750415802, + "kept_token_count": 573.2484375, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9793486893177032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.066936279647053e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.80594940185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.250390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8550, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3004625618457794, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.019320244342088698, + "grad_norm": 0.13152560591697693, + "kept_token_count": 572.35625, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9782078087329864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.512602071277797e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.87986221313477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.61896057128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8560, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.3009075552225113, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.020962400082498788, + "grad_norm": 0.08510733395814896, + "kept_token_count": 572.11875, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9788245379924774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.271685073850677e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.08411712646485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.00184936523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8570, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2997437447309494, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.018195544276386498, + "grad_norm": 0.12239161133766174, + "kept_token_count": 573.1515625, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9791711091995239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.136307005770504e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.85597152709961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6482849121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8580, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.30119627714157104, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.016449570748955013, + "grad_norm": 0.1537882536649704, + "kept_token_count": 571.8625, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9797897040843964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.894669179222546e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.57890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2273345947266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2994964629411697, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.016331489384174346, + "grad_norm": 0.12389489263296127, + "kept_token_count": 573.36875, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9783006846904755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.476316870655864e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.06938018798829, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.19085083007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8600, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.30074198842048644, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.01873952466994524, + "grad_norm": 0.12200578302145004, + "kept_token_count": 572.4234375, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9782694160938263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.48853298521135e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.7368392944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.72926025390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8610, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.30005123317241666, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.016715306136757136, + "grad_norm": 0.10980021953582764, + "kept_token_count": 572.7484375, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9782536089420318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.494706271449104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.1854461669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.43495483398436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8620, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30086706578731537, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.01856259647756815, + "grad_norm": 0.1290191262960434, + "kept_token_count": 572.09375, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.979061895608902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.178967254934832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.14971923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.98493041992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8630, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29968788027763366, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.01846557855606079, + "grad_norm": 0.13097158074378967, + "kept_token_count": 573.0734375, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.979003083705902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.201940945582465e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.8134880065918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9863708496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8640, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.2994149476289749, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01734509812667966, + "grad_norm": 0.11241424083709717, + "kept_token_count": 573.1859375, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9796178340911865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.96179760072846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.68447799682617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.84388122558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29971546232700347, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.016536460630595683, + "grad_norm": 0.1426541805267334, + "kept_token_count": 572.9796875, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9790301084518432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.191384622477927e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.48095169067383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3156280517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8660, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30031167566776273, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.018581430427730083, + "grad_norm": 0.12459800392389297, + "kept_token_count": 572.409375, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9781953930854798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.51744473038707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.8121940612793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5767028808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8670, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.300658655166626, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.01903729299083352, + "grad_norm": 0.14255501329898834, + "kept_token_count": 572.1921875, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9788924872875213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.245141216320918e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.04816055297852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1002716064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8680, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30073236525058744, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.016115658078342678, + "grad_norm": 0.11630655825138092, + "kept_token_count": 572.009375, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9786712408065796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.33156969747506e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.50815658569336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1448638916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8690, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.30072673261165617, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.015982757229357956, + "grad_norm": 0.11283896118402481, + "kept_token_count": 572.2421875, + "learning_rate": 2.551973050585119e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9792678236961365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.098523103399202e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.11862258911133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.12852172851564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.30109919905662536, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.019886675756424665, + "grad_norm": 0.1580154150724411, + "kept_token_count": 571.7828125, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9774088382720947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.824687101878226e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.40887603759765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.26746826171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8710, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.30032913088798524, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.017136968951672316, + "grad_norm": 0.12374652177095413, + "kept_token_count": 572.6625, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9777737140655518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.682158077135682e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44426879882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.81153259277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8720, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.29988616704940796, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.01625129198655486, + "grad_norm": 0.1509172022342682, + "kept_token_count": 573.203125, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9788147389888764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.27551368274726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.33334732055664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2516632080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8730, + "valid_token_count": 818.7296875 + }, + { + "actual_token_mask_ratio": 0.3002599090337753, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.01707559172064066, + "grad_norm": 0.11888465285301208, + "kept_token_count": 572.596875, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9799265027046203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.841240731067955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.66761779785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8000457763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8740, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.29932396709918974, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.015277249738574028, + "grad_norm": 0.13383343815803528, + "kept_token_count": 573.23125, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9783882677555085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.442098260275088e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.17487258911133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1423400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8750, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2993918865919113, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.01572849852964282, + "grad_norm": 0.12977488338947296, + "kept_token_count": 573.4765625, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9794789671897888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.016047868295573e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.36284255981445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.24375915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8760, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3004456788301468, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.019358379859477283, + "grad_norm": 0.09095518290996552, + "kept_token_count": 572.296875, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.978844404220581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.263919880846515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.00910568237305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5509399414062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8770, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30000199377536774, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.017993869539350272, + "grad_norm": 0.15954501926898956, + "kept_token_count": 572.90625, + "learning_rate": 2.542502846342621e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9784568548202515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.415307092946023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.95399627685546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4207702636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8780, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2996252626180649, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.017028173618018626, + "grad_norm": 0.11614525318145752, + "kept_token_count": 573.0796875, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9787024676799774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.319369953824207e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.43005142211913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.88184814453126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8790, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.299496191740036, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.014025484211742878, + "grad_norm": 0.12913252413272858, + "kept_token_count": 573.2578125, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9787802219390869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.288995959446766e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.96361541748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3988067626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2998447507619858, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.015414133481681348, + "grad_norm": 0.12387160211801529, + "kept_token_count": 573.0109375, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9760229051113128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 9.366074300487526e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.9204345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.76508178710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8810, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30013352632522583, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.017212485149502754, + "grad_norm": 0.11694274097681046, + "kept_token_count": 572.7375, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9777647316455841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.685671418788843e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.87616348266602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.35094299316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8820, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.29898928105831146, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.01849723793566227, + "grad_norm": 0.12165788561105728, + "kept_token_count": 573.725, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9780131220817566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.588643977418542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.84928588867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.5172637939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.29981099963188174, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.01987837441265583, + "grad_norm": 0.10375846922397614, + "kept_token_count": 572.9453125, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9779215276241302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.62442517245654e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.44377746582032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.44464111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8840, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29948248863220217, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.01851518955081701, + "grad_norm": 0.09679348021745682, + "kept_token_count": 573.1765625, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9776693999767303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.722907150513493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.96309814453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7275054931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8850, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30014601051807405, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.017154920659959317, + "grad_norm": 0.11963161826133728, + "kept_token_count": 572.6890625, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9785074472427369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.3955455920659e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.42113723754883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0241455078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8860, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30004594624042513, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.019524112343788147, + "grad_norm": 0.13520723581314087, + "kept_token_count": 572.71875, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.977325564622879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.857224383973517e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.01250991821288, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9045867919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8870, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3012107044458389, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.016030512191355227, + "grad_norm": 0.12358183413743973, + "kept_token_count": 571.86875, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9772152721881866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.900310931494459e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.52439880371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7073699951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.29968676865100863, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.018603702168911697, + "grad_norm": 0.10524676740169525, + "kept_token_count": 573.1125, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9790566742420197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.18100517790299e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.92679748535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3172332763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8890, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.3000792026519775, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.017129215504974128, + "grad_norm": 0.1011732965707779, + "kept_token_count": 572.7390625, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.978062903881073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.569200508645736e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.00398712158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.32354431152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8900, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3005355894565582, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.016201428696513177, + "grad_norm": 0.11446963995695114, + "kept_token_count": 572.44375, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9792255342006684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.115048476611264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.69734344482421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1326538085938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3018741339445114, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.017439591605216263, + "grad_norm": 0.1168450191617012, + "kept_token_count": 571.1046875, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9785506188869476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.37869090901222e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.18532791137696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2087066650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8920, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.2996789664030075, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.021207192353904246, + "grad_norm": 0.11079537123441696, + "kept_token_count": 573.0078125, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9785275995731354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.387674315599725e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.1410285949707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3746795654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8930, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.30067821145057677, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.01873670406639576, + "grad_norm": 0.1244584396481514, + "kept_token_count": 572.3859375, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9787851870059967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.287057062261738e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.7448486328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1876586914062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8940, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.30011754631996157, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.01640021661296487, + "grad_norm": 0.09927414357662201, + "kept_token_count": 572.9078125, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9794712483882904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.019064480322414e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.05561676025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5189239501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8950, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.30041069686412813, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.018814815487712622, + "grad_norm": 0.11695799976587296, + "kept_token_count": 572.359375, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9791344285011292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.1506377318874e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.04088668823242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.71039123535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8960, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29986563622951506, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.01969635458663106, + "grad_norm": 0.1295582354068756, + "kept_token_count": 572.7953125, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.978971540927887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.214261106331833e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.79691619873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0503387451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8970, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2990287184715271, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.017350288573652506, + "grad_norm": 0.13625089824199677, + "kept_token_count": 573.471875, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9788644492626191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.256087749032303e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4266586303711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.44724731445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8980, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2995084375143051, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.018120749108493327, + "grad_norm": 0.12831233441829681, + "kept_token_count": 573.2609375, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9797776997089386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.899355550762266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.94651870727539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.61651000976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8990, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29964197874069215, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.0173577681183815, + "grad_norm": 0.10548695921897888, + "kept_token_count": 573.0203125, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9802425384521485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.717771586612798e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.28926544189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.82005615234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9000, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2993840128183365, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.018644431419670583, + "grad_norm": 0.22662653028964996, + "kept_token_count": 573.5125, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9778793156147003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.640909800305963e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.84205169677735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.0805725097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9010, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.3001720428466797, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.018415796011686324, + "grad_norm": 0.13936184346675873, + "kept_token_count": 572.61875, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9786694943904877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.332249126397073e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.76418914794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3891357421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9020, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30108446478843687, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.018567738775163888, + "grad_norm": 0.1175377368927002, + "kept_token_count": 571.94375, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9773899674415588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.832067032926716e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.99187393188477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7385955810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9030, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.29967111349105835, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.016774755157530307, + "grad_norm": 0.100022092461586, + "kept_token_count": 573.0234375, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9780129373073578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.588711280026473e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.36829071044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.99267578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9040, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2996930718421936, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.015800803992897273, + "grad_norm": 0.10863696783781052, + "kept_token_count": 573.3, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9782970905303955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.477719238726422e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.80828247070312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.71636352539065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9050, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.3007426768541336, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.017883892077952622, + "grad_norm": 0.11468227207660675, + "kept_token_count": 572.3890625, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9784762442111969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.407735076616518e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.69910049438477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.62969360351565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9060, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30031668245792387, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.01841725306585431, + "grad_norm": 0.115404412150383, + "kept_token_count": 572.5265625, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9782043159008026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.513959401170723e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.99994354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0009307861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9070, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30010542571544646, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.017121550999581812, + "grad_norm": 0.11617596447467804, + "kept_token_count": 572.9625, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9793158113956452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.079776016529649e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.96628112792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2659637451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9080, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.30024253726005556, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.017331458535045384, + "grad_norm": 0.15518486499786377, + "kept_token_count": 572.7296875, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9783039450645447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.475042050122283e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.65589218139648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7038970947266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9090, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.3000383049249649, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.017624662816524507, + "grad_norm": 0.09811670333147049, + "kept_token_count": 572.778125, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9784461259841919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.41950204630848e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.45967178344726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6876678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9100, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.3000437796115875, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.015736465621739627, + "grad_norm": 0.13195167481899261, + "kept_token_count": 572.7984375, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9786806464195251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.32788682600949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.47558135986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3825317382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3007287114858627, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.017607400473207236, + "grad_norm": 0.10454767197370529, + "kept_token_count": 572.178125, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9785194575786591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.390858565689995e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.54241714477538, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7270141601563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9120, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3008897632360458, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.016321033332496883, + "grad_norm": 0.09596506506204605, + "kept_token_count": 572.16875, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9798542439937592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.869455876061693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.06722106933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.39299926757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9130, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2999486804008484, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.016245646588504314, + "grad_norm": 0.09076583385467529, + "kept_token_count": 572.9140625, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9787454307079315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.302588685182855e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.02607498168945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9865020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9140, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2996627360582352, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.019987561739981174, + "grad_norm": 0.1380225569009781, + "kept_token_count": 573.128125, + "learning_rate": 2.497660336787154e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9777782082557678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.68039860506542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07348251342773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6838684082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9150, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29894435703754424, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.01596920480951667, + "grad_norm": 0.10010609030723572, + "kept_token_count": 573.61875, + "learning_rate": 2.496425005357775e-05, + "loss": 0.016, + "mgd_cos_sim": 0.977932333946228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.620198714197614e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.33679962158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.41981506347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9160, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.299745973944664, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.017982370406389236, + "grad_norm": 0.14854243397712708, + "kept_token_count": 572.971875, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9791740357875824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.135159223456866e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01160583496093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.83848876953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9170, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3009145587682724, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.016481386870145796, + "grad_norm": 0.0879804715514183, + "kept_token_count": 571.8359375, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9781546652317047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.533354266546667e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.31322860717773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.470751953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9180, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.2991106122732162, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.01867880066856742, + "grad_norm": 0.23626425862312317, + "kept_token_count": 573.628125, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9791754901409149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.134594900184311e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.08679885864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9699462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9190, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.30027511715888977, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.01780208358541131, + "grad_norm": 0.10296700149774551, + "kept_token_count": 572.496875, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9788252711296082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.271395490737632e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76672286987305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0814575195312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29913762509822844, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.018707990925759078, + "grad_norm": 0.12895940244197845, + "kept_token_count": 573.3703125, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9788247644901276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.271593760582619e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.89154357910157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2280334472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9210, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.3001939028501511, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.017923513147979974, + "grad_norm": 0.128217875957489, + "kept_token_count": 572.571875, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9781018137931824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.554003361496143e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9443344116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7632629394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9220, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.29993531107902527, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.018073306139558553, + "grad_norm": 0.10589504987001419, + "kept_token_count": 572.98125, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9776190638542175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.742570134927519e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.10874786376954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.27100830078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9230, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.29925097823143004, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.01611432069912553, + "grad_norm": 0.13771575689315796, + "kept_token_count": 573.459375, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9790640890598297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.178109201253391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.65543746948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.40488891601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9240, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2993412882089615, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.018913279753178358, + "grad_norm": 0.1376716047525406, + "kept_token_count": 573.3953125, + "learning_rate": 2.485252719338089e-05, + "loss": 0.019, + "mgd_cos_sim": 0.978409343957901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.433869516011328e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.06664199829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4566192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9250, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2992497444152832, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.01824005814269185, + "grad_norm": 0.1155511662364006, + "kept_token_count": 573.328125, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9796111285686493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.964418546180241e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.6484001159668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.24310302734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9260, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.30008222460746764, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.02075655907392502, + "grad_norm": 0.1309918761253357, + "kept_token_count": 573.0671875, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9785973966121674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.36041261209175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23088302612305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5192077636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9270, + "valid_token_count": 818.765625 + }, + { + "actual_token_mask_ratio": 0.3004299610853195, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.01858661314472556, + "grad_norm": 0.1626499444246292, + "kept_token_count": 572.603125, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9787152528762817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.314375299960375e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.66308288574218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3114440917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9280, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2996469378471375, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.016081646736711263, + "grad_norm": 0.12104755640029907, + "kept_token_count": 572.9984375, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9791446030139923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.146664476953447e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.38641815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5089630126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9290, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2988789141178131, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.018209964968264103, + "grad_norm": 0.1268036663532257, + "kept_token_count": 573.6375, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9793750643730164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.05663425126113e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.05471572875976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4218994140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9300, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30039978623390196, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.016328195855021477, + "grad_norm": 0.14204730093479156, + "kept_token_count": 572.2265625, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9775662660598755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.763196310610511e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.46633987426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.96815795898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9310, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.30017822086811063, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.01551230875775218, + "grad_norm": 0.12611371278762817, + "kept_token_count": 572.8046875, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9789304494857788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.230314197135157e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.71339340209961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.21514587402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9320, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.3001708835363388, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.017130804155021905, + "grad_norm": 0.1412455141544342, + "kept_token_count": 572.678125, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9776378750801087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.735223163967021e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.99846038818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.67466735839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9330, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2993904709815979, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.021544204372912646, + "grad_norm": 0.10519207268953323, + "kept_token_count": 573.3390625, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9785426795482636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.381781226489693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.8003173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.09837951660154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9340, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3002658635377884, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.017566481605172157, + "grad_norm": 0.09742499887943268, + "kept_token_count": 572.6109375, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9791807472705841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.13254329841584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.71745452880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.95457763671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3001402199268341, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.016741100791841747, + "grad_norm": 0.13012056052684784, + "kept_token_count": 572.7421875, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9789400398731232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.226562931668013e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.74739379882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.22142333984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.30023460984230044, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.01573374103754759, + "grad_norm": 0.11168347299098969, + "kept_token_count": 572.59375, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9789952516555787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.20499990368262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.33874206542968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4981292724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2998435527086258, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014374783728271723, + "grad_norm": 0.09362015873193741, + "kept_token_count": 572.8921875, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9804075062274933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.653341308468952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.39433212280274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0965637207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30130836367607117, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.017665090411901473, + "grad_norm": 0.1388799101114273, + "kept_token_count": 571.8546875, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9782964944839477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.477951632812619e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.97727279663086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3583709716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9390, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.300517800450325, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.01779988091439009, + "grad_norm": 0.11949435621500015, + "kept_token_count": 572.4015625, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9788003444671631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.281131958938203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.94694595336914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.60328979492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9400, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.3003301441669464, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.01850115731358528, + "grad_norm": 0.14169779419898987, + "kept_token_count": 572.6859375, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9781521320343017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.534346998203545e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.1218276977539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8993713378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9410, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.3008485347032547, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.01497213114053011, + "grad_norm": 0.11401510238647461, + "kept_token_count": 572.1296875, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9802703320980072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.70691542129498e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.55126647949218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1852966308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9420, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2997241079807281, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.01592186987400055, + "grad_norm": 0.12220262736082077, + "kept_token_count": 573.3296875, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9808686196804046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.473215446225367e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.6914077758789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.36170959472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9430, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.2997283130884171, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.01642323173582554, + "grad_norm": 0.1159777119755745, + "kept_token_count": 573.028125, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9791652143001557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.13860839116387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.07483749389648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.83070373535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9440, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.3006085455417633, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.01750757321715355, + "grad_norm": 0.16918852925300598, + "kept_token_count": 572.0953125, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9791098892688751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.160218421835453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.52156372070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.75718078613284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9450, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.2996625751256943, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.018335996475070716, + "grad_norm": 0.1498875617980957, + "kept_token_count": 573.0109375, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9776013910770416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.749479748075828e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.51799621582032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7162780761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9460, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.3007423341274261, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.01836499935016036, + "grad_norm": 0.10870254039764404, + "kept_token_count": 572.365625, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9783547997474671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.455175193375908e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5060531616211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8171142578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9470, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.300095808506012, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.016938939318060876, + "grad_norm": 0.11542493104934692, + "kept_token_count": 572.6265625, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9787538111209869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.299311011796817e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.463818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.48809814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9480, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3004061311483383, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.016943167243152858, + "grad_norm": 0.11034256219863892, + "kept_token_count": 572.3296875, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9791428327560425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.147355110850185e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.06394424438477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.33128662109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9490, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2997245639562607, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.01451287530362606, + "grad_norm": 0.10190092772245407, + "kept_token_count": 572.990625, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9786159455776214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.353168013854883e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.5750518798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.81080322265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9500, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3005047470331192, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.01730685429647565, + "grad_norm": 0.11835427582263947, + "kept_token_count": 572.4390625, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9797026157379151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.928682243800722e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.47512512207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0161071777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9510, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29904082119464875, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.01871169190853834, + "grad_norm": 0.10671058297157288, + "kept_token_count": 573.2609375, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9786137402057647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.35402162920218e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.15362777709962, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.52293090820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9520, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.2992209941148758, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.013984220940619707, + "grad_norm": 0.08893676102161407, + "kept_token_count": 573.428125, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9789562463760376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.22023837827146e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.8578987121582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.65679626464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9530, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.300309881567955, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.01684429831802845, + "grad_norm": 0.09722435474395752, + "kept_token_count": 572.6109375, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9790061175823211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.200754018616863e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.73263854980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.0624237060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9540, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.3001255840063095, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.016226619016379118, + "grad_norm": 0.10893339663743973, + "kept_token_count": 572.7796875, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9798971176147461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.85270647611469e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.77582092285157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.51328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9550, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.2998927295207977, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.014681640453636647, + "grad_norm": 0.08089431375265121, + "kept_token_count": 572.9625, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9795968472957611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.969994403538294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.16349716186524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3319458007812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9560, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.29946318864822385, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.017265330627560617, + "grad_norm": 0.12268128246068954, + "kept_token_count": 573.0625, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.979122668504715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.155225877999328e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.40336685180664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.71441040039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9570, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.30024216175079343, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.01815532213076949, + "grad_norm": 0.13018424808979034, + "kept_token_count": 572.859375, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9789080262184143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.239067465183325e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.21591720581054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.97667236328124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9580, + "valid_token_count": 818.64375 + }, + { + "actual_token_mask_ratio": 0.3000999420881271, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.017879070527851582, + "grad_norm": 0.15562061965465546, + "kept_token_count": 572.6390625, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9782434344291687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.498677925672382e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.57742385864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7524139404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2995665013790131, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017598468344658613, + "grad_norm": 0.1233067661523819, + "kept_token_count": 573.2515625, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9793731391429901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.057388622546568e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.49613342285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4911285400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9600, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.30018098950386046, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.015735494345426558, + "grad_norm": 0.13649199903011322, + "kept_token_count": 572.4859375, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9782161474227905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.509337276336737e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.11412658691407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.96673889160155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9610, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.3003271073102951, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.01687572617083788, + "grad_norm": 0.09724526107311249, + "kept_token_count": 572.5, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9791515707969666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.143937739077956e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.10075912475585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8968200683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9620, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2995649755001068, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.01640375144779682, + "grad_norm": 0.0995737686753273, + "kept_token_count": 573.125, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9792238235473633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.115714517771266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.20426483154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.11102294921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9630, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30057071447372435, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.017004031874239444, + "grad_norm": 0.1321616917848587, + "kept_token_count": 572.2328125, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9797705113887787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.902166180429049e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.9087127685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.396044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9640, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.3001653403043747, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.016632997989654542, + "grad_norm": 0.11561750620603561, + "kept_token_count": 572.7609375, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9789194285869598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.234617926063947e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76551361083985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9558990478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9650, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2999151170253754, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.015106456726789475, + "grad_norm": 0.1290280818939209, + "kept_token_count": 573.0671875, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.978756308555603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.29833050374873e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.49917831420899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7246948242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9660, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.3007410794496536, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.015782291162759067, + "grad_norm": 0.11043322831392288, + "kept_token_count": 572.2296875, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9791015982627869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.163460224750452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.33165130615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.1971801757812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3003163903951645, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015094356704503297, + "grad_norm": 0.09213905781507492, + "kept_token_count": 572.5390625, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9797782182693482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.899152842583135e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.20629959106445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3729553222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9680, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30012080669403074, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.015952611807733774, + "grad_norm": 0.09499593824148178, + "kept_token_count": 572.5625, + "learning_rate": 2.429257583115679e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9793814063072205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.054156860453077e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.29331741333007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5105926513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9690, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3011352300643921, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.018022308126091957, + "grad_norm": 0.11511217057704926, + "kept_token_count": 571.8734375, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9784783661365509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.406913402723149e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61200256347657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.576806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9700, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2997604787349701, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.017116528656333684, + "grad_norm": 0.1354425847530365, + "kept_token_count": 572.9984375, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9781465351581573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.536527820979245e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.34884719848633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7391784667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9710, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30000644624233247, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.017017382476478816, + "grad_norm": 0.10044043511152267, + "kept_token_count": 572.9, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9797240495681763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.92031132732518e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.37541351318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.95673217773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9720, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29997970461845397, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.014250515028834343, + "grad_norm": 0.10678699612617493, + "kept_token_count": 572.8375, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9794119238853455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.04223891464062e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.17012329101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4642822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9730, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3005490243434906, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.018984621576964855, + "grad_norm": 0.12960046529769897, + "kept_token_count": 572.2109375, + "learning_rate": 2.422753693407244e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9798539757728577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.869553664932027e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9266975402832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.7216369628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9740, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.3004589080810547, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.018173164594918488, + "grad_norm": 0.12677150964736938, + "kept_token_count": 572.43125, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9798080623149872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.887491738074459e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.45407409667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.7410522460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9750, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30000136196613314, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.016934278327971695, + "grad_norm": 0.12235084176063538, + "kept_token_count": 572.675, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9783464670181274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.458433148916811e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.1855224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9135803222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9760, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29987933337688444, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.016105848457664253, + "grad_norm": 0.08306833356618881, + "kept_token_count": 573.053125, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9791491806507111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.144869862007909e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.95811538696289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8605255126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9770, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.29957843720912936, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.019048095773905517, + "grad_norm": 0.1206841841340065, + "kept_token_count": 572.9484375, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9786784708499908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.328745898324996e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.48210906982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5087951660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9780, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.2992836326360703, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.01886776741594076, + "grad_norm": 0.1013740599155426, + "kept_token_count": 573.4578125, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9785923361778259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.362388180103154e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.23505477905273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5021057128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9790, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.29908539056777955, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.015192743670195342, + "grad_norm": 0.09554651379585266, + "kept_token_count": 573.703125, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9793762147426606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.056186125031673e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.54311599731446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.48858642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9800, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.3003456234931946, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.01557887876406312, + "grad_norm": 0.10508470237255096, + "kept_token_count": 572.3, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9782154440879822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.509616600349545e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.73680801391602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4977783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9810, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.2996376663446426, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.01527700237929821, + "grad_norm": 0.10954160988330841, + "kept_token_count": 573.0796875, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9786002397537231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.359301064047031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.8620391845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.19536743164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9820, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.29994170665740966, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.016698007192462682, + "grad_norm": 0.14262628555297852, + "kept_token_count": 572.85, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9799355030059814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.837713637854904e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.42173919677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2734680175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2985030829906464, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.01653706477954984, + "grad_norm": 0.10743298381567001, + "kept_token_count": 573.8640625, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9802293181419373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.722939553787001e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.52692565917968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.68219604492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9840, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.3001808226108551, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.017925665993243457, + "grad_norm": 0.15560057759284973, + "kept_token_count": 572.3390625, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9800283730030059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.801437823218293e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.34464492797852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.19853515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9850, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.30013037025928496, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.017457649018615483, + "grad_norm": 0.11888563632965088, + "kept_token_count": 572.4734375, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9795535981655121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.986899654497393e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.9569580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.62819519042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9860, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.29998519122600553, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.017047572787851094, + "grad_norm": 0.10261649638414383, + "kept_token_count": 573.175, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9791295886039734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.152521622832865e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.39606628417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2084136962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9870, + "valid_token_count": 818.803125 + }, + { + "actual_token_mask_ratio": 0.29996755719184875, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.017068801447749138, + "grad_norm": 0.12352852523326874, + "kept_token_count": 572.8125, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9794274687767028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.036163853830658e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.05346984863282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.39610290527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9880, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2997527033090591, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.017705017887055873, + "grad_norm": 0.1461952030658722, + "kept_token_count": 573.0234375, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9787429094314575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.30356206279248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.4329963684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.1394989013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9890, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3003211170434952, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.016497265174984933, + "grad_norm": 0.1250481903553009, + "kept_token_count": 572.4796875, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9795097947120667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.004006449482403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.73500289916993, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4551055908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9900, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2994009256362915, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.01647826572880149, + "grad_norm": 0.1454780250787735, + "kept_token_count": 573.4734375, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9803145825862885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.689631529501639e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.24873275756836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3906646728516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9910, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.30026443004608155, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.019887467194348574, + "grad_norm": 0.12122444063425064, + "kept_token_count": 572.4265625, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.979689770936966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.933697634143755e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.78330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8618682861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9920, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.30072005093097687, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.019401259440928697, + "grad_norm": 0.15095333755016327, + "kept_token_count": 572.2328125, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9789567708969116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.220034942496568e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.05894927978515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0436187744141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9930, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29954189658164976, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.01660090507939458, + "grad_norm": 0.11358915269374847, + "kept_token_count": 573.284375, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9802210986614227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.726145922788419e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.79824981689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.50908813476565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9940, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2983731716871262, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.01653171218931675, + "grad_norm": 0.11657790094614029, + "kept_token_count": 573.990625, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9797561585903167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.907767576398328e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.70907440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.08363952636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.3004987746477127, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01567189684137702, + "grad_norm": 0.09695205837488174, + "kept_token_count": 572.315625, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9792363941669464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.110800117719919e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.02269592285157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.80674438476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9960, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29983026087284087, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.01578268697485328, + "grad_norm": 0.11756576597690582, + "kept_token_count": 573.1046875, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9791688740253448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.137177646858618e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.28049011230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0685150146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9970, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.3007389694452286, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.01689610779285431, + "grad_norm": 0.12319771945476532, + "kept_token_count": 572.415625, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9797959685325622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.892220673966222e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.13421936035157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.859912109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9980, + "valid_token_count": 818.603125 + }, + { + "actual_token_mask_ratio": 0.29973678290843964, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.01509395856410265, + "grad_norm": 0.11781862378120422, + "kept_token_count": 573.00625, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9799463748931885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.833461786503904e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.3152847290039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.71400146484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9990, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3004750370979309, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.016562712471932174, + "grad_norm": 0.09387840330600739, + "kept_token_count": 572.2984375, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9804965496063233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.618556555826217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.1239028930664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.71649780273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10000, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.3002373903989792, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.014931273553520441, + "grad_norm": 0.10866554826498032, + "kept_token_count": 572.7671875, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9796780169010162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.938294656923972e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.534375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5170196533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10010, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.2988483101129532, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.01741869319230318, + "grad_norm": 0.10363183915615082, + "kept_token_count": 573.78125, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9796776056289673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.938454364193603e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.1495361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1333770751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10020, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3002473533153534, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.017367587331682442, + "grad_norm": 0.11798693984746933, + "kept_token_count": 572.85, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9801101684570312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.769489311613143e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.95977325439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.5049285888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10030, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.299712136387825, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.016044017486274244, + "grad_norm": 0.11011666059494019, + "kept_token_count": 572.9484375, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9804350852966308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.642558848601766e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.56489486694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.7909790039063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3004055380821228, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.014820696692913771, + "grad_norm": 0.10234268754720688, + "kept_token_count": 572.3546875, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9798032164573669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.889385669841431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.26754837036133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7411224365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.30034900903701783, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.015853399597108363, + "grad_norm": 0.0979064330458641, + "kept_token_count": 572.55625, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9782883524894714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.481130935251713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.9416893005371, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.12354431152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10060, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29903280138969424, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.016243684757500886, + "grad_norm": 0.13218843936920166, + "kept_token_count": 573.6234375, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9788907468318939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.245825520134532e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.09542312622071, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5780456542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10070, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2989163130521774, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.016818679962307216, + "grad_norm": 0.10793700814247131, + "kept_token_count": 573.49375, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.978828901052475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.269977952295449e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.74293060302735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.13307495117186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10080, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.2997791886329651, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.015980696585029364, + "grad_norm": 0.09064430743455887, + "kept_token_count": 572.909375, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9796161949634552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.962443269207142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.5129638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6068603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10090, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2997150182723999, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.017820219881832598, + "grad_norm": 0.10615891963243484, + "kept_token_count": 573.1171875, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9797580301761627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.907039616839029e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5058807373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2541931152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2999724805355072, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.01755061959847808, + "grad_norm": 0.12506623566150665, + "kept_token_count": 572.625, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9787800073623657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.28907883260399e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.79591827392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.66344299316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10110, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.29949136078357697, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.016184152103960514, + "grad_norm": 0.15177197754383087, + "kept_token_count": 573.2703125, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9795812666416168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.976086635608226e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72961044311523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2701477050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10120, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2998106449842453, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.015033998619765043, + "grad_norm": 0.11757934838533401, + "kept_token_count": 572.7578125, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9805005669593811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.616985094500705e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27570877075195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.82245178222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10130, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.2997915357351303, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.016782677639275788, + "grad_norm": 0.15057441592216492, + "kept_token_count": 573.29375, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9795642256736755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.982742172316649e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.23567810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.32607421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10140, + "valid_token_count": 818.7421875 + }, + { + "actual_token_mask_ratio": 0.3005096107721329, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.013896974548697472, + "grad_norm": 0.090445876121521, + "kept_token_count": 572.4, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9798161029815674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.884358346927911e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.7473876953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6353729248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10150, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3013851284980774, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.015427227690815925, + "grad_norm": 0.12532779574394226, + "kept_token_count": 571.609375, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9795761168003082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.978100620675832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.66725616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.01820068359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10160, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29975993037223814, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.019119726959615946, + "grad_norm": 0.11513902246952057, + "kept_token_count": 573.096875, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9800491273403168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.793329423293472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22706832885743, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9304931640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10170, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2991961628198624, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.01574922064319253, + "grad_norm": 0.11763686686754227, + "kept_token_count": 573.43125, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9783289492130279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.465269929729402e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.57505340576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8506164550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10180, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3004736125469208, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.014306750241667032, + "grad_norm": 0.10221083462238312, + "kept_token_count": 572.375, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9801525712013245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.752920282655396e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.83979034423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5733642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10190, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.30081126391887664, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01593861049041152, + "grad_norm": 0.1552591472864151, + "kept_token_count": 572.334375, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9798299491405487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.878952237660997e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.91826400756835, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8796661376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10200, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.3002021938562393, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.015925728622823952, + "grad_norm": 0.0964091569185257, + "kept_token_count": 572.6875, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.016, + "mgd_cos_sim": 0.979825633764267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.880623306846246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.55462646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.983935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10210, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.2993353813886642, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.01826501442119479, + "grad_norm": 0.12058955430984497, + "kept_token_count": 573.309375, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9797703921794891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.902212091721594e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.80314483642579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.687109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10220, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2999859720468521, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.015618200553581119, + "grad_norm": 0.07002104073762894, + "kept_token_count": 572.6046875, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9804104924201965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.652169879293069e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.17373504638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.34229431152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10230, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.2987584710121155, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.015216978639364243, + "grad_norm": 0.10101020336151123, + "kept_token_count": 573.89375, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9797004401683808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.929535859148018e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.91736526489258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.01621704101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3008843719959259, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.014691117499023677, + "grad_norm": 0.10269221663475037, + "kept_token_count": 571.9875, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9789409458637237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.226210193242877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.81560974121093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8532958984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10250, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.29979070127010343, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.017322100419551133, + "grad_norm": 0.0902240127325058, + "kept_token_count": 572.91875, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9803219914436341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.686743629164994e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44255218505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6631256103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10260, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2997764855623245, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.016999422945082186, + "grad_norm": 0.13213883340358734, + "kept_token_count": 572.9609375, + "learning_rate": 2.352128522412946e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9801055431365967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.771288583171553e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.01434936523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.30694885253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10270, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.300702965259552, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.017398027703166008, + "grad_norm": 0.11296819895505905, + "kept_token_count": 571.978125, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9797473967075347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.91119055065792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.60948257446289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7701843261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10280, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.30053991377353667, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.01569619569927454, + "grad_norm": 0.11479784548282623, + "kept_token_count": 572.228125, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9802511513233185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.714409512118436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.54382781982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4947113037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10290, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.29976366758346557, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.01709633832797408, + "grad_norm": 0.10796697437763214, + "kept_token_count": 573.0265625, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9799660503864288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.825779175618663e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.66957092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8719940185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10300, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3002407312393188, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.017107066046446563, + "grad_norm": 0.11833649128675461, + "kept_token_count": 572.465625, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9787916123867035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.284545328933746e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.53036804199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9215087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10310, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.2998313337564468, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.014712585974484682, + "grad_norm": 0.10179977118968964, + "kept_token_count": 572.8046875, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9789227068424224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.23333532025572e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55180435180664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8407928466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10320, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.3006409019231796, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.016977111902087928, + "grad_norm": 0.10401709377765656, + "kept_token_count": 572.3640625, + "learning_rate": 2.343945413972411e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9799041211605072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.849972316762433e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.97220001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.61293029785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10330, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.300684654712677, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.015953862573951483, + "grad_norm": 0.13578170537948608, + "kept_token_count": 571.9296875, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9780575573444367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.57128827192355e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.74274291992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.369677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10340, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.301052588224411, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.016466278675943614, + "grad_norm": 0.10051551461219788, + "kept_token_count": 571.875, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9788126051425934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.276347871287726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.06011657714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.27842102050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10350, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30004689991474154, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.01717346431687474, + "grad_norm": 0.10597805678844452, + "kept_token_count": 572.9125, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9800606548786164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.788821967551485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.53710479736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.85776672363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10360, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.2992660850286484, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.015992092620581388, + "grad_norm": 0.07974434643983841, + "kept_token_count": 573.6125, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9801533401012421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.75262014940381e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.09265060424805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0765319824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10370, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.29979307651519777, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.016124302241951226, + "grad_norm": 0.10406240820884705, + "kept_token_count": 572.98125, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9783529579639435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.455895513179713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.05843582153321, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.16957397460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10380, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3006935328245163, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.018265822250396014, + "grad_norm": 0.11810246855020523, + "kept_token_count": 572.221875, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.979778128862381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.899185002315789e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.71406936645508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7566345214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10390, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3012943536043167, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.01646151253953576, + "grad_norm": 0.10082326084375381, + "kept_token_count": 572.0328125, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9799619734287262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.827373774489388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.28578872680664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.10386657714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10400, + "valid_token_count": 818.6984375 + }, + { + "actual_token_mask_ratio": 0.3014865696430206, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.016862375009804965, + "grad_norm": 0.11219743639230728, + "kept_token_count": 571.584375, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9794449925422668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.029317468754016e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.1278190612793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.79632263183595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10410, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2993736952543259, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.016755591705441476, + "grad_norm": 0.13969804346561432, + "kept_token_count": 573.140625, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9794006764888763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.046630391618237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.65440521240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5426422119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10420, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.29958653450012207, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014913493394851684, + "grad_norm": 0.08705416321754456, + "kept_token_count": 573.1578125, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9792919039726258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.08912343927659e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.95841827392579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4257537841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10430, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2995403170585632, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.015741212479770183, + "grad_norm": 0.12563510239124298, + "kept_token_count": 573.19375, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9803012073040008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.694856976740994e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.5593650817871, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6129425048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10440, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30034799575805665, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.016279933135956527, + "grad_norm": 0.1014866828918457, + "kept_token_count": 572.4, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9797292232513428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.918291157693602e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36280746459961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2379547119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10450, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2987836390733719, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.01722920723259449, + "grad_norm": 0.09940694272518158, + "kept_token_count": 573.625, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9814599037170411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.242244791996199e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.2113525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6335723876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10460, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.3001501619815826, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015826777555048466, + "grad_norm": 0.1311718076467514, + "kept_token_count": 572.675, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9801934242248536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.736966290394776e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.15164489746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.82342834472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10470, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.29966295659542086, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.015615086257457732, + "grad_norm": 0.11711151152849197, + "kept_token_count": 573.403125, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9799256980419159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.841540427762083e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.02750549316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1638641357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10480, + "valid_token_count": 818.753125 + }, + { + "actual_token_mask_ratio": 0.30030049085617067, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.016634144820272923, + "grad_norm": 0.09669490158557892, + "kept_token_count": 572.4984375, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9796385943889618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.953692911542021e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.6190284729004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2639404296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3005920767784119, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.019590338785201312, + "grad_norm": 0.11991869658231735, + "kept_token_count": 572.3140625, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9795109689235687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.003540788195095e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.58133239746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1197509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10500, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3006427675485611, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.014670887682586908, + "grad_norm": 0.10264339298009872, + "kept_token_count": 572.0890625, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9793562233448029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.063988861977123e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.94938201904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7579864501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10510, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.3010755926370621, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.015654023457318544, + "grad_norm": 0.11190950870513916, + "kept_token_count": 572.0046875, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9793561816215515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.06401323643513e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.87904205322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.40178833007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10520, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.300005179643631, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.018995868787169457, + "grad_norm": 0.12543900310993195, + "kept_token_count": 572.8109375, + "learning_rate": 2.316403786537544e-05, + "loss": 0.019, + "mgd_cos_sim": 0.979371988773346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.057831073529086e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.5150161743164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.67686157226564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10530, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2995137482881546, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.01703887516632676, + "grad_norm": 0.12079624831676483, + "kept_token_count": 573.0609375, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9797243118286133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.920210045995191e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.79918060302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.06077575683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10540, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.29995101392269136, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.017833564709872008, + "grad_norm": 0.11859811097383499, + "kept_token_count": 572.58125, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9786795258522034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.328330586664379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.3233543395996, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6261840820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10550, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.3013474941253662, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.015047577302902937, + "grad_norm": 0.11442504823207855, + "kept_token_count": 571.5625, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9802351772785187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.720653375145047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.5353759765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.66285400390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10560, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.300329852104187, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.016170857287943364, + "grad_norm": 0.10794070363044739, + "kept_token_count": 572.6921875, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9798987567424774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.85206604632549e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.04141006469726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10570, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.2993424475193024, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.015122808888554573, + "grad_norm": 0.08759970217943192, + "kept_token_count": 573.0796875, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9802481412887574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.715585015830584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.10257949829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.55213317871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10580, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.29937856197357177, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.016047308035194875, + "grad_norm": 0.09334366768598557, + "kept_token_count": 573.3, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9785781681537629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.367925911443309e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.12861862182618, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7411315917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10590, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2985075831413269, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.015695403702557086, + "grad_norm": 0.11365194618701935, + "kept_token_count": 574.2, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.97987459897995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.861506164772437e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.79207153320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6912445068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10600, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.29935279190540315, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.014386359043419361, + "grad_norm": 0.09135419875383377, + "kept_token_count": 573.5796875, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9806627452373504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.553634131909348e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.4026123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.883642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10610, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.29933984875679015, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.017174702044576405, + "grad_norm": 0.10506514459848404, + "kept_token_count": 573.3921875, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9798965632915497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.852922572055831e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.30756301879883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9884796142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10620, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.30110798478126527, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.01669711908325553, + "grad_norm": 0.09914200752973557, + "kept_token_count": 571.7171875, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9811993539333344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.344021796598099e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.47421798706054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.64837646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10630, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30006611049175264, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.01883156429976225, + "grad_norm": 0.09471127390861511, + "kept_token_count": 572.784375, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9804736614227295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.627495142514817e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.91190643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.43360290527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10640, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29992271661758424, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.016578942351043225, + "grad_norm": 0.10697860270738602, + "kept_token_count": 572.9109375, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9806370854377746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.563656981801614e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.07752914428711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4437591552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10650, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30061462223529817, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.012952232360839843, + "grad_norm": 0.10535340756177902, + "kept_token_count": 572.209375, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9806873261928558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.544026811956428e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.22352447509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0962707519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10660, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.2999201834201813, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.015613075997680425, + "grad_norm": 0.09813100099563599, + "kept_token_count": 572.6875, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9801677823066711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.746978189970832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.98670501708985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0665649414062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10670, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.3010630398988724, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.017438163980841637, + "grad_norm": 0.1486639380455017, + "kept_token_count": 571.8359375, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9789991557598114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.203474062611349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.58695678710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8354766845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10680, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.3015032529830933, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.01642777556553483, + "grad_norm": 0.11018797755241394, + "kept_token_count": 571.875, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9792851388454438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.091756899375468e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.26238174438477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0270935058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10690, + "valid_token_count": 818.728125 + }, + { + "actual_token_mask_ratio": 0.29939356744289397, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.01721691470593214, + "grad_norm": 0.10028810054063797, + "kept_token_count": 573.3953125, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9805907905101776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.581735626445152e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.07167129516601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.11626892089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10700, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.29993348121643065, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.016607664432376622, + "grad_norm": 0.11149094253778458, + "kept_token_count": 572.7546875, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9800860226154328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.778915096423589e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.90147399902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.14174194335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10710, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.3012054651975632, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.0181792956776917, + "grad_norm": 0.10765533149242401, + "kept_token_count": 571.5546875, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9799886524677277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.816956203896553e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.67430191040039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.97447509765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10720, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.2988485902547836, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.01681070923805237, + "grad_norm": 0.1365518569946289, + "kept_token_count": 573.5921875, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9800172805786133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.805766363162547e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.52222442626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.73267517089846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10730, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29942320585250853, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.017520715668797492, + "grad_norm": 0.10486700385808945, + "kept_token_count": 573.2015625, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9803956568241119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.65796889027115e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.37886199951171, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2415252685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10740, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.29929288029670714, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.017359765991568566, + "grad_norm": 0.12494093179702759, + "kept_token_count": 573.4015625, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9799746751785279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.822414045222104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.68837280273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.91114807128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10750, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.29947521090507506, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.016823259927332402, + "grad_norm": 0.11280114203691483, + "kept_token_count": 573.115625, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9796425104141235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.952167725306935e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7474464416504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.19908752441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10760, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.3006287306547165, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.01578339859843254, + "grad_norm": 0.1162627637386322, + "kept_token_count": 572.0453125, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9805084466934204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.613910565851256e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.51699752807617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7595458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10770, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.29917588233947756, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.013875157712027431, + "grad_norm": 0.07682917267084122, + "kept_token_count": 573.60625, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9800103604793549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.808470472809858e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.86571731567383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.45274963378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3003407210111618, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.015207597427070141, + "grad_norm": 0.1477842479944229, + "kept_token_count": 572.4609375, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9807008266448974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.53875880036503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.43928680419921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.37464904785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10790, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.29953100979328157, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.018255461007356644, + "grad_norm": 0.16062529385089874, + "kept_token_count": 573.2203125, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9778839468955993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.639103543828242e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.36947860717774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.23504333496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29935040771961213, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.015932121593505145, + "grad_norm": 0.09303989261388779, + "kept_token_count": 573.434375, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.016, + "mgd_cos_sim": 0.980522346496582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.608478335896507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.16559448242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4582458496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10810, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.3002215176820755, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.01648628283292055, + "grad_norm": 0.10922324657440186, + "kept_token_count": 572.575, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9796593725681305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.945571269374341e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9842903137207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1215393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10820, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30118913650512696, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.015296091418713331, + "grad_norm": 0.0980096310377121, + "kept_token_count": 572.05, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9797456383705139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.911876600701362e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.54534912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8649108886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10830, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.299773371219635, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.01593054858967662, + "grad_norm": 0.09553181380033493, + "kept_token_count": 573.053125, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9792897343635559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.0899651220534e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.15556488037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.97667236328124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10840, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30027020871639254, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.01736959395930171, + "grad_norm": 0.11542671918869019, + "kept_token_count": 572.690625, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.980616420507431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.571734677185305e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.24660110473633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.02434997558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10850, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29911785423755644, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.015916618146002294, + "grad_norm": 0.1222691535949707, + "kept_token_count": 573.61875, + "learning_rate": 2.270099049894723e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9796994686126709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.929917555884458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52614288330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4333526611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10860, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3001207411289215, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.016237314511090516, + "grad_norm": 0.09714154154062271, + "kept_token_count": 572.759375, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9794984340667725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.00844223704189e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.23462753295898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.68023376464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10870, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3003366678953171, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.016410044580698013, + "grad_norm": 0.12956203520298004, + "kept_token_count": 572.7109375, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.979999303817749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.812791300239041e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.78450393676758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.90265502929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10880, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.3007933974266052, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.01521206134930253, + "grad_norm": 0.09788268059492111, + "kept_token_count": 572.221875, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9799800157546997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.820326864020899e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.86824111938476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4509643554687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.30118730664253235, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.01623931610956788, + "grad_norm": 0.09365503489971161, + "kept_token_count": 571.7875, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9792368352413178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.110622584354133e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.0345085144043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.21666564941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10900, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3002408742904663, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.016128545626997948, + "grad_norm": 0.12027329206466675, + "kept_token_count": 572.484375, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9798695027828217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.863501523388549e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.62012939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9680969238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29937309622764585, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.015505286306142807, + "grad_norm": 0.10678926855325699, + "kept_token_count": 573.4796875, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9805949568748474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.580117453471757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4213363647461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4450286865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10920, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.29924132525920866, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.017191193904727696, + "grad_norm": 0.10323388874530792, + "kept_token_count": 573.4234375, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9787944078445434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.283448842121288e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.49203033447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5940399169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10930, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29988610446453096, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.015610936749726533, + "grad_norm": 0.11793754994869232, + "kept_token_count": 573.140625, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9804762780666352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.626474325661547e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.99779205322265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.82667846679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10940, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.29973346590995786, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.015132880304008722, + "grad_norm": 0.10564476996660233, + "kept_token_count": 573.103125, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.978528892993927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.387173220398836e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.79246139526367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4684692382813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10950, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.29912875294685365, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.015004499722272158, + "grad_norm": 0.10674335062503815, + "kept_token_count": 573.7109375, + "learning_rate": 2.255863302527758e-05, + "loss": 0.015, + "mgd_cos_sim": 0.97844757437706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.41893554024864e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.25267639160157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.45728454589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10960, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.2993695944547653, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.01869036993011832, + "grad_norm": 0.10262979567050934, + "kept_token_count": 573.3, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9803296446800231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.683749208808877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.51115036010742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.94085083007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10970, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29989573657512664, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.015249646920710801, + "grad_norm": 0.09572959691286087, + "kept_token_count": 572.8953125, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9799451470375061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.833948839106598e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.87290954589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8293487548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.3003554046154022, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.01794872935861349, + "grad_norm": 0.11220007389783859, + "kept_token_count": 572.5328125, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9796722412109375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.94054867583327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.17868576049804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.66839294433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30056250393390654, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.015429599862545728, + "grad_norm": 0.11959800869226456, + "kept_token_count": 572.3765625, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9804143309593201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.650674815522507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.33967666625976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.83457946777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11000, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29888194501399995, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.017388494778424503, + "grad_norm": 0.0879240408539772, + "kept_token_count": 573.653125, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9794309198856354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.034815109567717e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48991546630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8163665771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11010, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999336063861847, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.016241351235657932, + "grad_norm": 0.10850294679403305, + "kept_token_count": 572.9859375, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9800740897655487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.783577384543606e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.77289352416992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.992626953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11020, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.2999224692583084, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.015990398451685907, + "grad_norm": 0.10150633007287979, + "kept_token_count": 572.8578125, + "learning_rate": 2.245843529059e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9794576704502106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.024365452001803e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.58453750610352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3961517333984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11030, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2989990234375, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.01449234625324607, + "grad_norm": 0.10464039444923401, + "kept_token_count": 573.6171875, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9798058986663818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.888338950579055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.66735458374023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0057098388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11040, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.2988781094551086, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.016320823598653077, + "grad_norm": 0.09726820141077042, + "kept_token_count": 573.7046875, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9795003712177277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.00768582848832e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.65619201660157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2389831542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11050, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3004739671945572, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.016924979351460933, + "grad_norm": 0.08623545616865158, + "kept_token_count": 572.6046875, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9794351935386658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.033146223169751e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.34438858032226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.186669921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11060, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.29918335676193236, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.014746848493814468, + "grad_norm": 0.11548863351345062, + "kept_token_count": 573.390625, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9796110570430756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.964449032442645e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.23209075927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.82269897460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11070, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2997895866632462, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.016631822846829892, + "grad_norm": 0.13362625241279602, + "kept_token_count": 572.7953125, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9807425975799561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.522443193010985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.72353744506836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4808746337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11080, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.3001507431268692, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.015672725439071656, + "grad_norm": 0.10415662825107574, + "kept_token_count": 572.5875, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9790622651576996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.178820717148483e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.13241653442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.14599609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11090, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.29883733689785, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.015483798645436763, + "grad_norm": 0.09413977712392807, + "kept_token_count": 573.515625, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9808293282985687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.488562478101813e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.66668548583985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8458190917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11100, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.3006179004907608, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.016376958694308996, + "grad_norm": 0.10690529644489288, + "kept_token_count": 572.203125, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9796089947223663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.965256590978243e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.00303497314454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4348175048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11110, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.29956458806991576, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.01684608031064272, + "grad_norm": 0.1286378651857376, + "kept_token_count": 573.1140625, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9802162706851959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.728032796876505e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.84167861938477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.65184631347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11120, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29890632033348086, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.01864328971132636, + "grad_norm": 0.2422822117805481, + "kept_token_count": 573.45, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9779115140438079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.628333307569847e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.61299057006836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7367309570312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11130, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.29956434965133666, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.015703699458390473, + "grad_norm": 0.09428535401821136, + "kept_token_count": 573.01875, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9804406344890595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.640399926458485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.10770721435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6877136230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11140, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.3014989227056503, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.013968544639647007, + "grad_norm": 0.08387670665979385, + "kept_token_count": 571.5546875, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9805511116981507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.597240328323096e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.30271606445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.59340515136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11150, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30073691010475156, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.017472239024937154, + "grad_norm": 0.13764287531375885, + "kept_token_count": 572.1703125, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9797665774822235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.903701989562251e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.2030876159668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.10349426269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11160, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.3003507196903229, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.017296710051596165, + "grad_norm": 0.10227680206298828, + "kept_token_count": 572.2515625, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9797721743583679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.901509816292674e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.83934631347657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.51597290039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11170, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.30080512166023254, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.014761013630777597, + "grad_norm": 0.11248712986707687, + "kept_token_count": 572.09375, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9794358968734741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.032875848584808e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.32474746704102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2301300048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11180, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29975024461746214, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.018968122918158768, + "grad_norm": 0.09055963903665543, + "kept_token_count": 572.9921875, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9791454017162323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.146342079271563e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.1142677307129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.51694641113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11190, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3003917932510376, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.017055179830640553, + "grad_norm": 0.11907734721899033, + "kept_token_count": 572.4140625, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9791970551013947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.126172178890556e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.62260360717774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4484832763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3002120852470398, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.0172222962602973, + "grad_norm": 0.08935900032520294, + "kept_token_count": 572.4578125, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9803414285182953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.679152040509507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.0041290283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6762176513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11210, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.3004039466381073, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.014197124261409044, + "grad_norm": 0.10810156166553497, + "kept_token_count": 572.465625, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9805908262729645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.581722966278903e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.11362686157227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7256713867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11220, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30026785135269163, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.020088612288236617, + "grad_norm": 0.10869286209344864, + "kept_token_count": 572.2796875, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9790785551071167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.17245636426378e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.34752044677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.58232421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11230, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.3006889522075653, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.01610416332259774, + "grad_norm": 0.11194711178541183, + "kept_token_count": 572.053125, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9785079419612884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.395355034735985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.86313934326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.91631774902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11240, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.30019587874412534, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.01709947446361184, + "grad_norm": 0.11140498518943787, + "kept_token_count": 572.35625, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9787511765956879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.300341505673714e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.37985229492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.51332702636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11250, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.29936419129371644, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.01557912491261959, + "grad_norm": 0.09209511429071426, + "kept_token_count": 573.3921875, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9791579008102417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.141468642861583e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.00885696411133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.77308349609376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11260, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30028568804264066, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.015296013467013836, + "grad_norm": 0.09783852100372314, + "kept_token_count": 572.4828125, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9795769631862641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.97777138359379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.31970596313477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3855865478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11270, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.29975411891937254, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.01394344912841916, + "grad_norm": 0.1065518707036972, + "kept_token_count": 573.2390625, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9802854835987092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.700998758082278e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.66306381225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9169677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11280, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.29955093264579774, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.013593724882230162, + "grad_norm": 0.10858259350061417, + "kept_token_count": 573.2015625, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9808257639408111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.489954514312558e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.42269363403321, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3701110839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11290, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.29956123530864714, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.015082815289497375, + "grad_norm": 0.08553877472877502, + "kept_token_count": 573.2609375, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9796871662139892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.934722816571593e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.04512481689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.67189025878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11300, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2995390444993973, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.01610849155113101, + "grad_norm": 0.1212850883603096, + "kept_token_count": 573.4046875, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9793276786804199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.075143487076275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.7263313293457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4623718261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11310, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.3001006990671158, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.01503739720210433, + "grad_norm": 0.09710592031478882, + "kept_token_count": 572.8484375, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9800121247768402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.807779620634392e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.98464584350586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.88707580566404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11320, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.30044437646865846, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.015490684751421213, + "grad_norm": 0.08287862688302994, + "kept_token_count": 572.8609375, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9808283090591431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.488959599868394e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.21266708374023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2729064941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11330, + "valid_token_count": 818.890625 + }, + { + "actual_token_mask_ratio": 0.2998004585504532, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.017918483447283508, + "grad_norm": 0.10008556395769119, + "kept_token_count": 572.8109375, + "learning_rate": 2.200944709436783e-05, + "loss": 0.018, + "mgd_cos_sim": 0.978083199262619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.561270442442037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.3040771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.64268798828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11340, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.299666765332222, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.017057518288493157, + "grad_norm": 0.12340835481882095, + "kept_token_count": 573.0234375, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.979838103055954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.875759547459893e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.056201171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6163787841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11350, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2993845134973526, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.01704442100599408, + "grad_norm": 0.11406217515468597, + "kept_token_count": 573.2515625, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9788560569286346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.259367969003506e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9034523010254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6667205810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11360, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29921052753925326, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.01628165105357766, + "grad_norm": 0.08680965006351471, + "kept_token_count": 573.7984375, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9806908547878266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.542651583207771e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.39735870361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6314422607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11370, + "valid_token_count": 818.778125 + }, + { + "actual_token_mask_ratio": 0.2999073714017868, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.01627043876796961, + "grad_norm": 0.12462876737117767, + "kept_token_count": 572.984375, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9803099632263184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.69144629884977e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.82248992919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6353393554688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11380, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.3002681493759155, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.01799453478306532, + "grad_norm": 0.11085977405309677, + "kept_token_count": 572.6984375, + "learning_rate": 2.193624931121462e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9802878022193908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.700097412453034e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.41571426391602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3754943847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2999640077352524, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.016979243978857993, + "grad_norm": 0.13714346289634705, + "kept_token_count": 572.8890625, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9804867684841156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.622373159392737e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.26482315063477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.94383544921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11400, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.3007019102573395, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.017315967474132776, + "grad_norm": 0.099411241710186, + "kept_token_count": 572.4453125, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9791220963001251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.155446630553342e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.58057479858398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.05658569335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11410, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.3000158965587616, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.014676620811223983, + "grad_norm": 0.13774853944778442, + "kept_token_count": 572.740625, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9801663637161255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.747535055386834e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.66492004394532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5795928955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11420, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29962285459041593, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.014795462880283595, + "grad_norm": 0.09341452270746231, + "kept_token_count": 572.8765625, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9807566583156586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.51695431972621e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.03271255493163, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9715209960938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11430, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.3002887636423111, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.014841580018401145, + "grad_norm": 0.08092742413282394, + "kept_token_count": 572.7265625, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9806039512157441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.57659705413971e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.98602752685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1476196289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11440, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2991573989391327, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.01597375525161624, + "grad_norm": 0.09672287106513977, + "kept_token_count": 573.33125, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9798301219940185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.878878677729517e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.50135803222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1440399169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11450, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.29935785233974455, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.016552123054862023, + "grad_norm": 0.13424333930015564, + "kept_token_count": 573.3328125, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9787568390369416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.298127868329175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.37466659545899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.8886383056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2992528200149536, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.01437064316123724, + "grad_norm": 0.1127394586801529, + "kept_token_count": 573.515625, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9803948760032654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.658272588741965e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.20343170166015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.52240600585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11470, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.299629870057106, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.015543947461992502, + "grad_norm": 0.13097472488880157, + "kept_token_count": 572.825, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9794139325618744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.041448309086263e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.82423782348633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.89833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11480, + "valid_token_count": 817.8890625 + }, + { + "actual_token_mask_ratio": 0.2997423827648163, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.015804266091436148, + "grad_norm": 0.11401867121458054, + "kept_token_count": 573.0796875, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9804855644702911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.6228487159824e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.97789688110352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.785888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11490, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.29929372370243074, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.015509966667741537, + "grad_norm": 0.12336697429418564, + "kept_token_count": 573.3359375, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.979527622461319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.997048378456384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.05131683349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0880462646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11500, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3003763765096664, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.01658086720854044, + "grad_norm": 0.1511969417333603, + "kept_token_count": 572.4796875, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9800753593444824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.78308036387898e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.84965438842774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.37978210449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11510, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30050039291381836, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.018658107332885265, + "grad_norm": 0.09101219475269318, + "kept_token_count": 572.4890625, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9792848706245423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.091868876363151e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.20642852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.99431762695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11520, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.29955604672431946, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.018253665417432785, + "grad_norm": 0.11566434055566788, + "kept_token_count": 573.1625, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9800049602985382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.810581228113733e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.2066665649414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7476837158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11530, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30008252859115603, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.01839516470208764, + "grad_norm": 0.14519931375980377, + "kept_token_count": 572.81875, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9795646905899048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.982558236108162e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.67066192626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1442596435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11540, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2999603748321533, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.018050175905227662, + "grad_norm": 0.11307824403047562, + "kept_token_count": 572.859375, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9793066084384918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.083382490440272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.57190780639648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.91075744628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11550, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30062021911144254, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.016976843029260634, + "grad_norm": 0.09919103980064392, + "kept_token_count": 572.521875, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9798588514328003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.867655003792606e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.88046264648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.56474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11560, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.3005745679140091, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.015642416663467884, + "grad_norm": 0.1103687435388565, + "kept_token_count": 572.346875, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9797346949577331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.916149261291138e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.38046264648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.63140258789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11570, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.3004000723361969, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.017304698843508957, + "grad_norm": 0.08337537944316864, + "kept_token_count": 572.3890625, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9806258022785187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.568065739178564e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.80077056884765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2118255615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11580, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.30045951902866364, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.01779896942898631, + "grad_norm": 0.08573011308908463, + "kept_token_count": 572.265625, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9809384763240814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.445922237820923e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23343811035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.65575256347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11590, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.299535396695137, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.01589945163577795, + "grad_norm": 0.09826986491680145, + "kept_token_count": 573.3171875, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9794357001781464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.032949262997136e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34037704467774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1233703613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11600, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2997638463973999, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.01859251307323575, + "grad_norm": 0.11760521680116653, + "kept_token_count": 573.2375, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9791573226451874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.141690705087967e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52921752929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4485137939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11610, + "valid_token_count": 818.63125 + }, + { + "actual_token_mask_ratio": 0.2995868712663651, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.01918435366824269, + "grad_norm": 0.11287003755569458, + "kept_token_count": 573.1796875, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9791663944721222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.138147386489436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.48489913940429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6968139648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11620, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29973643720149995, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.019967226404696704, + "grad_norm": 0.1737271249294281, + "kept_token_count": 573.028125, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9797245264053345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.920120042399503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.89254531860351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.10552673339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.29925858974456787, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.01620944859459996, + "grad_norm": 0.13306233286857605, + "kept_token_count": 573.3015625, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9802989602088928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.69573736761231e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36729278564454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2617126464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11640, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29975667893886565, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.01632918920367956, + "grad_norm": 0.11206715553998947, + "kept_token_count": 573.13125, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9806062579154968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.575691706733778e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5359718322754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.81333923339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11650, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.29934623539447786, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.015347123146057129, + "grad_norm": 0.09556340426206589, + "kept_token_count": 573.35625, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9799806237220764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.820085957064294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.50942611694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1776916503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11660, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.2998930126428604, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.016191997472196816, + "grad_norm": 0.11402883380651474, + "kept_token_count": 572.8234375, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9806167542934418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.571597961941734e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.21315307617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6074523925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11670, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2998808443546295, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.015280600823462009, + "grad_norm": 0.10320916771888733, + "kept_token_count": 572.86875, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.97924103140831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.10900004580617e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.76056747436523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.49891662597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11680, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3003726631402969, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.015351783949881792, + "grad_norm": 0.13896021246910095, + "kept_token_count": 572.5328125, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9802114725112915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.729911521892063e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.16310119628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8934295654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2991951256990433, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.01802637819200754, + "grad_norm": 0.16084913909435272, + "kept_token_count": 573.396875, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9794266521930695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.036485596676357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.0360008239746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.92348327636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11700, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.30100647509098055, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.015923358034342527, + "grad_norm": 0.11027684807777405, + "kept_token_count": 571.759375, + "learning_rate": 2.146288936272548e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9796472191810608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.950321814860217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.24221572875976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.29563903808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11710, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.2994614034891129, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.01497633634135127, + "grad_norm": 0.10798335820436478, + "kept_token_count": 573.153125, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9808200895786285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.492167933378368e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.1752326965332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4194030761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11720, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2997983455657959, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.01743944687768817, + "grad_norm": 0.12103679776191711, + "kept_token_count": 573.0640625, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.979908573627472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.848229652154259e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.78491592407227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4182922363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11730, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.300245863199234, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.017132346983999014, + "grad_norm": 0.08426602929830551, + "kept_token_count": 572.6453125, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9800267875194549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.802053514751605e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.72897109985351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2950134277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11740, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2989154726266861, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.01824009222909808, + "grad_norm": 0.10749851912260056, + "kept_token_count": 573.6234375, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9792943894863129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.08814394986257e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.65622253417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7696197509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11750, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29957474768161774, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.017296543717384337, + "grad_norm": 0.1118256226181984, + "kept_token_count": 573.240625, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9795494258403779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.98852568550501e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.93376846313477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0135559082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11760, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3000587821006775, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.016331740841269492, + "grad_norm": 0.11072363704442978, + "kept_token_count": 572.709375, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9788404285907746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.265482829301618e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.99421310424805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1436828613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11770, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29965007305145264, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.013364448770880699, + "grad_norm": 0.10979700833559036, + "kept_token_count": 573.246875, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9800321221351623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.7999742643442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.24225311279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.62218322753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.3001164346933365, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.016485762596130372, + "grad_norm": 0.09958569705486298, + "kept_token_count": 572.7, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9792244493961334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.115463424474001e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.04016876220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.37425231933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11790, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.30065628588199617, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.017377869971096516, + "grad_norm": 0.11511125415563583, + "kept_token_count": 572.3734375, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9800622045993805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.7882208279334e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.09203414916992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6778503417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11800, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.30021131336688994, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.017043132334947586, + "grad_norm": 0.13441139459609985, + "kept_token_count": 572.64375, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9799437284469604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.834496427676641e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.20712509155274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.91981506347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11810, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.3006840318441391, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.016782056540250778, + "grad_norm": 0.12267206609249115, + "kept_token_count": 572.1515625, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.979488217830658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.012438411242329e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.72689895629883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.004833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11820, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.30002056062221527, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.01704392246901989, + "grad_norm": 0.10593412816524506, + "kept_token_count": 572.8734375, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9791376948356628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.1493489415152e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.57262802124023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1594268798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11830, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29960086941719055, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.01783236684277654, + "grad_norm": 0.12562818825244904, + "kept_token_count": 572.9515625, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9795790612697601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.976948472787626e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.66010284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3762481689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11840, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.29938014447689054, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.01644742414355278, + "grad_norm": 0.12451845407485962, + "kept_token_count": 573.3046875, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9791045427322388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.162300655385479e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.59745483398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.93475952148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11850, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3003205418586731, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.01531949508935213, + "grad_norm": 0.1120065450668335, + "kept_token_count": 572.575, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9801329135894775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.760600346955471e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99157943725587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5304229736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11860, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.29971186220645907, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.017195655964314938, + "grad_norm": 0.11635696142911911, + "kept_token_count": 573.2390625, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9804582893848419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.633497079950757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49351348876954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9531585693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11870, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.3005160421133041, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.016649658046662807, + "grad_norm": 0.22640123963356018, + "kept_token_count": 572.34375, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.980803382396698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.498698614654132e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10677947998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6377716064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11880, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3000268667936325, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.016499093919992446, + "grad_norm": 0.13491053879261017, + "kept_token_count": 572.853125, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9796531677246094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.948002457851544e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.16828918457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.61241149902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11890, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.30036213994026184, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.015939171053469182, + "grad_norm": 0.13439063727855682, + "kept_token_count": 572.5890625, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9800699591636658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.785185516695492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08384475708007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.91468811035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11900, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2999108225107193, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.01697431979700923, + "grad_norm": 0.1296805739402771, + "kept_token_count": 572.83125, + "learning_rate": 2.116291843968583e-05, + "loss": 0.017, + "mgd_cos_sim": 0.979347288608551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.067481394391507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.27695388793946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1486114501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11910, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29924688339233396, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.016476419009268283, + "grad_norm": 0.1177418977022171, + "kept_token_count": 573.45625, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9796511650085449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.948777783894911e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.33053894042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.54380493164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11920, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2994630217552185, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.019271104875952006, + "grad_norm": 0.12286533415317535, + "kept_token_count": 573.021875, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9806999742984772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.539089419879019e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.38965225219727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.58441772460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11930, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.30094010233879087, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.019251068588346243, + "grad_norm": 0.11906009912490845, + "kept_token_count": 572.0234375, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9799800336360931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.820323589839973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.21445846557617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3338104248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11940, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.30094931125640867, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.01607015887275338, + "grad_norm": 0.08837662637233734, + "kept_token_count": 572.1015625, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9805648863315582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.59186157665681e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.62435760498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.05565185546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.30043603777885436, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.017893316969275475, + "grad_norm": 0.11907987296581268, + "kept_token_count": 572.3328125, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9793047249317169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.084110304480418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.6324562072754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1883117675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11960, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.29982813000679015, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.015591379813849926, + "grad_norm": 0.09105002135038376, + "kept_token_count": 573.0546875, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9802671849727631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.708143530180678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.46425247192383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8348785400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11970, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.30061326920986176, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.016446658223867417, + "grad_norm": 0.14484427869319916, + "kept_token_count": 572.5109375, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9796939074993134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.93208739196416e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.50401077270507, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2680145263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11980, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.3000487148761749, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.014703689888119698, + "grad_norm": 0.11727990210056305, + "kept_token_count": 572.621875, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9804513871669769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.636201116838493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.06522216796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1701995849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11990, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3004543036222458, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.0164040083065629, + "grad_norm": 0.09634940326213837, + "kept_token_count": 572.5171875, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.980825012922287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.490244242944755e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.3005584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2356689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12000, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3019290596246719, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.01807267628610134, + "grad_norm": 0.1337594985961914, + "kept_token_count": 571.13125, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9802159607410431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.728155542281456e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.79104461669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7012969970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12010, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30126773416996, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.013972612656652927, + "grad_norm": 0.14400199055671692, + "kept_token_count": 571.746875, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9800262689590454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.802260370226577e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.77624740600587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5846740722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12020, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3000759303569794, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.015516265202313662, + "grad_norm": 0.10250350087881088, + "kept_token_count": 572.4734375, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9800253033638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.802628279023338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.72419128417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.19178161621096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12030, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.3000799953937531, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.015504919085651636, + "grad_norm": 0.0992903932929039, + "kept_token_count": 572.653125, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9806794822216034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.547098066424951e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.46557083129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.379541015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2996348559856415, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.016534757800400257, + "grad_norm": 0.10411285609006882, + "kept_token_count": 573.1796875, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9809972584247589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.422973067150451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.46363067626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.06435241699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2992256969213486, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.017621511500328778, + "grad_norm": 0.11348274350166321, + "kept_token_count": 573.5484375, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9795167624950409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.00128240371123e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.68627319335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3480987548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12060, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3004664868116379, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.0146528541110456, + "grad_norm": 0.2260715812444687, + "kept_token_count": 572.278125, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9790379464626312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.188317078747787e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.70392990112305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0501190185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12070, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.2998585432767868, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.015233395993709565, + "grad_norm": 0.09746064990758896, + "kept_token_count": 573.0578125, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9795830309391022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.975399057613686e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77160949707032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0935516357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12080, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.3009417146444321, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.014265487622469663, + "grad_norm": 0.0884915366768837, + "kept_token_count": 572.065625, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9798055469989777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.888481923146173e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.40743560791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.86851806640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12090, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29899649918079374, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.01645982451736927, + "grad_norm": 0.10113392770290375, + "kept_token_count": 573.6296875, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.978600925207138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.359027779079043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.64745559692383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.78953552246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12100, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29964258074760436, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.01634820308536291, + "grad_norm": 0.09945409744977951, + "kept_token_count": 572.8390625, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9801429033279419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.756694976706058e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.24681015014649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2164703369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12110, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.29949310421943665, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.015695608220994473, + "grad_norm": 0.13186132907867432, + "kept_token_count": 573.365625, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.980945485830307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.443187132594176e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.15684280395507, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2261260986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12120, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.29965661466121674, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.016559689771384, + "grad_norm": 0.17142212390899658, + "kept_token_count": 573.1796875, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9804286003112793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.64509430155158e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.89551467895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0147705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12130, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.300528347492218, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.01590065909549594, + "grad_norm": 0.11993816494941711, + "kept_token_count": 572.3953125, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9795232772827148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.998739092727192e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.45836486816407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4125640869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12140, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2997024029493332, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.015309532452374697, + "grad_norm": 0.08894390612840652, + "kept_token_count": 573.04375, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.98037890791893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.66451186791528e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.31245651245118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.24672546386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12150, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3003588318824768, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.016566750034689902, + "grad_norm": 0.1250050961971283, + "kept_token_count": 572.3953125, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9802285194396972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.723248418187723e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.46010284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7595611572266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12160, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3008242607116699, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.015064111165702342, + "grad_norm": 0.1130618005990982, + "kept_token_count": 571.9390625, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9796623587608337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.944409298943356e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.32004318237304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9682220458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12170, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.300029057264328, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.01806527879089117, + "grad_norm": 0.10343517363071442, + "kept_token_count": 572.825, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9801148056983948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.767671559122391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.37412643432617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.69428100585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12180, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29999615252017975, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.014624074101448059, + "grad_norm": 0.18590453267097473, + "kept_token_count": 572.8625, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9786905229091645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.324037262354977e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.30459899902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.88726806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12190, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.3006435513496399, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.016888607386499645, + "grad_norm": 0.11575862765312195, + "kept_token_count": 572.2046875, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9797470569610596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.911325010354631e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.86736602783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.82684631347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12200, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.299884295463562, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.018856240343302487, + "grad_norm": 0.16392819583415985, + "kept_token_count": 572.8328125, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9792517185211181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.104820008156822e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42406005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.79535522460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12210, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3003612220287323, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.014925201330333948, + "grad_norm": 0.13570627570152283, + "kept_token_count": 572.509375, + "learning_rate": 2.069209735969855e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9798122107982635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.88587196439039e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.05601577758789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6444976806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12220, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.3002526044845581, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.016590577736496925, + "grad_norm": 0.11590465903282166, + "kept_token_count": 572.5515625, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9795194566249847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.000230809557252e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.9445198059082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.08335876464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12230, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2989234119653702, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.016946111526340246, + "grad_norm": 0.12629635632038116, + "kept_token_count": 573.575, + "learning_rate": 2.066148765527252e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9798673272132874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.864344661356881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.17644882202148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.29344482421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12240, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2994948625564575, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.014411254972219466, + "grad_norm": 0.09771715849637985, + "kept_token_count": 573.2359375, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9805103480815888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.61316972784698e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.05009613037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.16331176757814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30038489401340485, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.016618183441460134, + "grad_norm": 0.12986843287944794, + "kept_token_count": 572.4921875, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9785799026489258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.367241607629695e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.55036163330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8190673828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12260, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.299610635638237, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.018238554522395135, + "grad_norm": 0.0966842994093895, + "kept_token_count": 573.35, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9801320850849151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.76092114392668e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.97944869995118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.13092651367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12270, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.2989208251237869, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.014054279960691928, + "grad_norm": 0.10499659925699234, + "kept_token_count": 573.9515625, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9804278433322906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.645392179256305e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.66010055541992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.74765930175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12280, + "valid_token_count": 818.6703125 + }, + { + "actual_token_mask_ratio": 0.3006758809089661, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013511515688151121, + "grad_norm": 0.0940946489572525, + "kept_token_count": 572.434375, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9805998861789703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.578188815386966e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.74817504882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.17084350585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12290, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.30054360032081606, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.018221574369817972, + "grad_norm": 0.14933712780475616, + "kept_token_count": 572.5375, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9801631450653077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.748786301817746e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.11658477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4547637939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12300, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2995952934026718, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.01676801182329655, + "grad_norm": 0.11839836090803146, + "kept_token_count": 573.1078125, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9798394322395325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.87523778853938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42315444946288, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2018737792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12310, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2997483849525452, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.01611863076686859, + "grad_norm": 0.09894832223653793, + "kept_token_count": 572.796875, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.980493426322937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.619770549354144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.27630233764648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.72740783691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12320, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.29824152290821077, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.014626000635325908, + "grad_norm": 0.204584538936615, + "kept_token_count": 574.4765625, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9805838108062744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.58447109546978e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.1500473022461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.07127990722654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12330, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.2998869359493256, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.015692160185426473, + "grad_norm": 0.11062635481357574, + "kept_token_count": 572.775, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9803806066513061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.663844371563755e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.87210464477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.88294982910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12340, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30065802335739134, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.01733978921547532, + "grad_norm": 0.09077004343271255, + "kept_token_count": 572.209375, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.979810893535614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.886387902544811e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.42577209472657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0779266357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.300084725022316, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.016053812578320503, + "grad_norm": 0.0983305349946022, + "kept_token_count": 572.9984375, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9792296707630157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.113427247735672e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.82388763427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0033477783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12360, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.2989909052848816, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.014612172078341246, + "grad_norm": 0.09055145829916, + "kept_token_count": 573.6765625, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9800276756286621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.801712927175685e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.80242614746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5977478027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12370, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30001832246780397, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.014325565379112958, + "grad_norm": 0.09082730114459991, + "kept_token_count": 572.9828125, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9805670440196991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.591018074890599e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.62335433959962, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.13836364746095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12380, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30059194266796113, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.015522858779877424, + "grad_norm": 0.10071288049221039, + "kept_token_count": 572.3125, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9802129745483399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.729327844572254e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5395263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8233917236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12390, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3006098806858063, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.015033265482634306, + "grad_norm": 0.14092057943344116, + "kept_token_count": 572.5859375, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.980864304304123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.474899612134323e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.55813522338867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9786071777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12400, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.2999127507209778, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.013953852374106646, + "grad_norm": 0.10270624607801437, + "kept_token_count": 573.1734375, + "learning_rate": 2.04002092226413e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9806319057941437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.565677115053404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.45248413085938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4318145751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12410, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.29918144047260287, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.014345105178654194, + "grad_norm": 0.10960214585065842, + "kept_token_count": 573.3578125, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9811960995197296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.345293561229482e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.40845031738282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.87425537109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12420, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30007428824901583, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.018347852863371372, + "grad_norm": 0.11559926718473434, + "kept_token_count": 572.5984375, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9795622110366822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.983530886122026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.35543212890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.26117553710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12430, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.30091145634651184, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.013703223736956715, + "grad_norm": 0.11559981107711792, + "kept_token_count": 572.2921875, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9810678124427795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.395404682029039e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.63213729858398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.39154663085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12440, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.29881325662136077, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.014650773629546166, + "grad_norm": 0.11919747292995453, + "kept_token_count": 573.8859375, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9802352547645569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.720616631559096e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.97821807861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5173034667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12450, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.301125231385231, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.013413951825350523, + "grad_norm": 0.07418365031480789, + "kept_token_count": 572.00625, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9805598080158233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.59384987759404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.91361389160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0028411865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12460, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2987990856170654, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.017087408807128668, + "grad_norm": 0.11542617529630661, + "kept_token_count": 573.8828125, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9800798654556274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.781321837683209e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.35476837158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.83465576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12470, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30012755692005155, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.014022372290492057, + "grad_norm": 0.1015206053853035, + "kept_token_count": 572.5375, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.979615980386734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.962528834468686e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.41585159301758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.73320922851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12480, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.30035367906093596, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016666808072477578, + "grad_norm": 0.10999827831983566, + "kept_token_count": 572.53125, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9802193641662598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.726832118351012e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.25720977783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.72515869140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12490, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30035487115383147, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.015108302515000106, + "grad_norm": 0.12183193862438202, + "kept_token_count": 572.4328125, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9800490260124206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.793373733875342e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.91909561157226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.65244140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12500, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2994450777769089, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.016813511215150356, + "grad_norm": 0.12507639825344086, + "kept_token_count": 573.4484375, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9798909962177277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.855095464037731e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.34111175537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.61402893066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12510, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3004027813673019, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.013040759414434434, + "grad_norm": 0.09031609445810318, + "kept_token_count": 572.4, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9803900301456452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.660164847038687e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.05663757324218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8683197021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12520, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.29939022064208987, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.016180241759866477, + "grad_norm": 0.13016077876091003, + "kept_token_count": 573.3921875, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9802526116371155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.713835220783949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.78439331054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9770935058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12530, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.29948224127292633, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.014512189943343402, + "grad_norm": 0.1352122277021408, + "kept_token_count": 573.2125, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9802813827991486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.702597067691386e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.74113616943359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.69568786621096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.29914192855358124, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.01566352238878608, + "grad_norm": 0.10399354249238968, + "kept_token_count": 573.35, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.980964869260788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.435616935254074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.10034561157227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4769744873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12550, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.30042328834533694, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.018187789153307675, + "grad_norm": 0.10698895156383514, + "kept_token_count": 572.2828125, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9811497330665588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.36340727598872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.82538375854492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.81189880371096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12560, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.3009191393852234, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.015384558960795403, + "grad_norm": 0.13246963918209076, + "kept_token_count": 572.040625, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9803148090839386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.689551566727459e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.64376220703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2577301025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12570, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.29990939497947694, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.01718167820945382, + "grad_norm": 0.12873448431491852, + "kept_token_count": 572.925, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.981033843755722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.40867129934486e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.58204498291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8096130371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12580, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30017776787281036, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.01508847689256072, + "grad_norm": 0.09744703024625778, + "kept_token_count": 572.69375, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9795671820640564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.981585076777264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55854263305665, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6625183105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30042983293533326, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.013660680223256349, + "grad_norm": 0.09565537422895432, + "kept_token_count": 572.49375, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9803580105304718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.672670762985945e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82994079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4132904052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12600, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2990332096815109, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.013113084342330694, + "grad_norm": 0.07631094753742218, + "kept_token_count": 573.5203125, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9790891230106353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.16832711279858e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.37660751342773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.437890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12610, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.30024886727333067, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.016392578091472387, + "grad_norm": 0.12241862714290619, + "kept_token_count": 572.546875, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9798297762870789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.879009208409116e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.33150711059571, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7997650146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12620, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.3009972482919693, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.01651677843183279, + "grad_norm": 0.09469854831695557, + "kept_token_count": 571.81875, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.980861222743988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.476107712136581e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.39036712646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4805969238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.29896360337734224, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.017046598438173532, + "grad_norm": 0.11702556908130646, + "kept_token_count": 573.5015625, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9806865870952606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.544320178567431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9021614074707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5898864746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12640, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.3003905177116394, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.014859552960842848, + "grad_norm": 0.09692519903182983, + "kept_token_count": 572.3390625, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9805217087268829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.608727246406488e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.1107276916504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3069122314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12650, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2993659347295761, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.01559996921569109, + "grad_norm": 0.09305068105459213, + "kept_token_count": 573.3046875, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9808822512626648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.467893374268897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.24592895507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.23711547851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.29876789450645447, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.01613007942214608, + "grad_norm": 0.1056574210524559, + "kept_token_count": 573.69375, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9800325036048889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.799818413332105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.1203598022461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.32220153808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12670, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.29980299770832064, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.015472621098160744, + "grad_norm": 0.09269680827856064, + "kept_token_count": 573.21875, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9811307251453399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.370834646280855e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.85787963867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1081298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12680, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.3004689455032349, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.016905372170731424, + "grad_norm": 0.10935993492603302, + "kept_token_count": 572.6625, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9809348404407501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.44734876207076e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.20406951904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3835388183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12690, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.3009232759475708, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.014591764658689499, + "grad_norm": 0.10163482278585434, + "kept_token_count": 571.9578125, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9805018305778503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.616487055202014e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.04427566528321, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5572875976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12700, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30059307515621186, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.01576658170670271, + "grad_norm": 0.09633059054613113, + "kept_token_count": 572.4015625, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9813060522079468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.30234089132864e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.6724723815918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4327392578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12710, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30247622132301333, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.01744410051032901, + "grad_norm": 0.10136391222476959, + "kept_token_count": 570.6203125, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9806544780731201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.556862401543185e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.95559539794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.77402648925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12720, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.3005413293838501, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.017861070204526187, + "grad_norm": 0.11405356228351593, + "kept_token_count": 572.234375, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9807206690311432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.531007213401608e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.8845588684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6285003662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12730, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30004661679267886, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.01435536341741681, + "grad_norm": 0.0877445787191391, + "kept_token_count": 572.765625, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9809089124202728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.457476531271822e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.93663787841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.59070739746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12740, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.299447163939476, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.01547458441928029, + "grad_norm": 0.0952923446893692, + "kept_token_count": 573.1125, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9809695243835449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.43380092899315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.52989959716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.50771484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12750, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.300622895359993, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.01658203606493771, + "grad_norm": 0.09142228960990906, + "kept_token_count": 572.375, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9809249103069305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.451223500538617e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.4714584350586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1636505126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12760, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3000191956758499, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.017632116470485924, + "grad_norm": 0.12037338316440582, + "kept_token_count": 572.74375, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.980750459432602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.519371647504158e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.32880020141602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.42148742675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12770, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29880746006965636, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.015910391043871642, + "grad_norm": 0.13471800088882446, + "kept_token_count": 573.71875, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9806657910346985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.552443858003244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.8452980041504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6183319091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12780, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.29797827899456025, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.017071467731148003, + "grad_norm": 0.1019715890288353, + "kept_token_count": 574.30625, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9786707103252411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.33176996820839e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.11821594238282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5051971435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12790, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30093005001544954, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.01538758808746934, + "grad_norm": 0.10853441059589386, + "kept_token_count": 571.7921875, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.980966055393219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.435158659063745e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.04292678833008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8235748291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12800, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.30083917677402494, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.014280549902468919, + "grad_norm": 0.09657308459281921, + "kept_token_count": 572.134375, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9806059122085571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.575831623398698e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01635284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.88136901855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12810, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2987295001745224, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.015910117886960506, + "grad_norm": 0.0738411396741867, + "kept_token_count": 574.01875, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9799166738986969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.845064319553786e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.32330551147462, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3605194091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12820, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.30005972683429716, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.01580246314406395, + "grad_norm": 0.11006568372249603, + "kept_token_count": 573.0171875, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9792432904243469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.10811121482402e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.10668029785157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7765869140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12830, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.3006352841854095, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015519177168607711, + "grad_norm": 0.11315059661865234, + "kept_token_count": 572.3796875, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9801229119300843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.764506226521916e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.73206176757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.63561096191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12840, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.3002759963274002, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.01751921707764268, + "grad_norm": 0.12367722392082214, + "kept_token_count": 572.690625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9795721530914306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.979643705766649e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.86405639648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.06458435058596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12850, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2998509883880615, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.015235487092286348, + "grad_norm": 0.10675721615552902, + "kept_token_count": 572.978125, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9805857360363006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.583717015222647e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.95581359863282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.75177001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12860, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2995315283536911, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.016043852549046278, + "grad_norm": 0.23717591166496277, + "kept_token_count": 573.078125, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9807370245456696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.524614338763058e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.08320693969726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.57619323730466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12870, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29832789599895476, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.014597366377711297, + "grad_norm": 0.10071371495723724, + "kept_token_count": 574.0015625, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9804619431495667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.63206699048169e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.57838516235351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2449066162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12880, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.29989380240440366, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.016801305767148732, + "grad_norm": 0.15332788228988647, + "kept_token_count": 572.7609375, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9810515284538269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.401762559311464e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25007781982421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3377410888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12890, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.3004226118326187, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.01381492828950286, + "grad_norm": 0.112979955971241, + "kept_token_count": 572.425, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.980754041671753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.517971025663427e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.6341323852539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9496307373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.30020058453083037, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.014959028642624616, + "grad_norm": 0.11339225620031357, + "kept_token_count": 572.6328125, + "learning_rate": 1.962109974767429e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9808870315551758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.46602214348968e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.65805740356446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.35601806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30025256872177125, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.016427973844110967, + "grad_norm": 0.12036903202533722, + "kept_token_count": 572.55625, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9811760246753692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.353137480095028e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.2642593383789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.56529541015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12920, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.298812472820282, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.01770902667194605, + "grad_norm": 0.12132672965526581, + "kept_token_count": 573.8484375, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9799714088439941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.82368726504501e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.66935272216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5186370849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12930, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3009339809417725, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.015353505499660969, + "grad_norm": 0.12573082745075226, + "kept_token_count": 572.125, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.979411917924881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.042242843657732e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.91083450317383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5165771484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12940, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2993704825639725, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.016415566112846135, + "grad_norm": 0.11344162374734879, + "kept_token_count": 573.5078125, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9795520782470704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.987489007064142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.90430374145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.85442504882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12950, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.3007645934820175, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.016070450749248268, + "grad_norm": 0.1100466400384903, + "kept_token_count": 572.1265625, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9795194506645203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.000235320650972e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.0592658996582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4469970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12960, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2989831119775772, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.01410510465502739, + "grad_norm": 0.09878502041101456, + "kept_token_count": 573.4234375, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9803613305091858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.671380444662645e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.07167282104493, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2819763183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12970, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.29963307082653046, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.015010959748178721, + "grad_norm": 0.10971767455339432, + "kept_token_count": 573.178125, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9801177382469177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.766524722683244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.14730606079101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.40198669433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12980, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29951665103435515, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.01522633396089077, + "grad_norm": 0.1349983960390091, + "kept_token_count": 573.2171875, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9806888997554779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.543418614659458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.07451477050782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.45354919433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.29892379641532896, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.01766681745648384, + "grad_norm": 0.10932693630456924, + "kept_token_count": 573.615625, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9797987520694733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.891129207564518e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44598007202148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8513580322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13000, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2993874788284302, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.015451772417873145, + "grad_norm": 0.13782784342765808, + "kept_token_count": 573.371875, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9801238179206848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.764152323943562e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49712219238282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.288037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13010, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2986845105886459, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.01591245997697115, + "grad_norm": 0.10612490028142929, + "kept_token_count": 573.95625, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9797588646411896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.906713217380457e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5753402709961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.115625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13020, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2992861449718475, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.015035598818212748, + "grad_norm": 0.13293439149856567, + "kept_token_count": 573.365625, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9814584016799927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.242828432936221e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.74795684814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.32521362304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13030, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3004318714141846, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.014587067626416683, + "grad_norm": 0.0863313227891922, + "kept_token_count": 572.3453125, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9810118496417999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.41726478736382e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.72973556518555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.40936584472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13040, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2991083115339279, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.01693854369223118, + "grad_norm": 0.15723884105682373, + "kept_token_count": 573.5984375, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9811060428619385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.380470633506775e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.74785614013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.44042053222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13050, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.29929222762584684, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.014843934215605259, + "grad_norm": 0.11741243302822113, + "kept_token_count": 573.3359375, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9817979574203491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.110187943908386e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.80389633178712, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4416229248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13060, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2996779590845108, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.013444627914577723, + "grad_norm": 0.10257595032453537, + "kept_token_count": 573.240625, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9799364864826202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.837320590624586e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.58400955200196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.11997375488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13070, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29967857599258424, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.01688268454745412, + "grad_norm": 0.140326127409935, + "kept_token_count": 573.0984375, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.980253529548645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.713478262303398e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.38328628540039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3907989501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13080, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29977707862854003, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.01678436202928424, + "grad_norm": 0.14436738193035126, + "kept_token_count": 573.325, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9815893948078156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.191666372818872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.73157577514648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9744506835938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13090, + "valid_token_count": 818.7703125 + }, + { + "actual_token_mask_ratio": 0.3004073679447174, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.014615086186677218, + "grad_norm": 0.1495237648487091, + "kept_token_count": 572.409375, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9805127739906311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.612216650159098e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.87260665893555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.47032470703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13100, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29989070296287534, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.017276996467262506, + "grad_norm": 0.12645211815834045, + "kept_token_count": 572.76875, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9801019251346588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.772705357638188e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.56848754882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.35621337890626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13110, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2997684061527252, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.014105648268014193, + "grad_norm": 0.10773880779743195, + "kept_token_count": 573.071875, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9815054655075073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.224444489111193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.68862075805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3596618652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13120, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30053685903549193, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.016211694199591876, + "grad_norm": 0.10714954882860184, + "kept_token_count": 572.446875, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9802360892295837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.720295834587887e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.4482650756836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.32277221679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13130, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3007803618907928, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.014251655247062445, + "grad_norm": 0.10166514664888382, + "kept_token_count": 572.20625, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.980978810787201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.430167606798932e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.79709548950196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.85113830566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30044485330581666, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.015536585729569197, + "grad_norm": 0.09545517712831497, + "kept_token_count": 572.390625, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9806500256061554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.558602665085346e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.5239387512207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6180908203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13150, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30025657415390017, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.013982639741152525, + "grad_norm": 0.07569222897291183, + "kept_token_count": 572.703125, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9805566430091858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.595079296152107e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.8862403869629, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2365966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13160, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.30105532705783844, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.01661716178059578, + "grad_norm": 0.08621355891227722, + "kept_token_count": 572.03125, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9803812861442566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.663581272936426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.4505485534668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.663916015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13170, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.2998282492160797, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.015236739162355661, + "grad_norm": 0.09797027707099915, + "kept_token_count": 572.8, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9804663002490998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.630368709214963e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.42378540039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9992431640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13180, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.30032719373703004, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.017411967925727367, + "grad_norm": 0.10328724980354309, + "kept_token_count": 572.428125, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9798077762126922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.887601823313162e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44646530151367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.24375915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13190, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2998230904340744, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.014074021484702826, + "grad_norm": 0.09808844327926636, + "kept_token_count": 572.74375, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9802492737770081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.715144747635349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.82494354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7530242919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13200, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.30037967562675477, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.01578198643401265, + "grad_norm": 0.09193333238363266, + "kept_token_count": 572.45625, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9807853460311889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.505735484301113e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.10771026611329, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.923193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13210, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.29999018609523775, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.01512132454663515, + "grad_norm": 0.09484636783599854, + "kept_token_count": 572.49375, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9803169012069702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.688728728680872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7575912475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5504608154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13220, + "valid_token_count": 817.846875 + }, + { + "actual_token_mask_ratio": 0.2993631064891815, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.015713702701032162, + "grad_norm": 0.09228697419166565, + "kept_token_count": 573.35625, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9808720827102662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.471858480130322e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.0997444152832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6254486083984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13230, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29954556524753573, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.013696210738271476, + "grad_norm": 0.09020599722862244, + "kept_token_count": 573.0109375, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9803583979606628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.672514766454697e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.57429809570313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.32778015136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13240, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.2998565465211868, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.015675637125968932, + "grad_norm": 0.09425409883260727, + "kept_token_count": 572.9234375, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9803469836711883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.67697682022117e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.21298904418946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.15928649902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13250, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2992294430732727, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.0173777861520648, + "grad_norm": 0.18876567482948303, + "kept_token_count": 573.66875, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9810530543327332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.401164912153035e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.94522323608399, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2373870849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13260, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.3014345794916153, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.01500282478518784, + "grad_norm": 0.0849691852927208, + "kept_token_count": 571.1625, + "learning_rate": 1.905131250213355e-05, + "loss": 0.015, + "mgd_cos_sim": 0.98074671626091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.520829531131313e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.73555755615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9480255126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13270, + "valid_token_count": 817.621875 + }, + { + "actual_token_mask_ratio": 0.29933707118034364, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.015020875353366136, + "grad_norm": 0.09337390214204788, + "kept_token_count": 573.509375, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9803971230983735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.657390015083365e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.45198822021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.643310546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13280, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.2997914254665375, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.016506389155983923, + "grad_norm": 0.12208328396081924, + "kept_token_count": 572.9140625, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9801579058170319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.750834702164866e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.69994583129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.11317749023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13290, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29949577450752257, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.014441921934485435, + "grad_norm": 0.1312672197818756, + "kept_token_count": 573.08125, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9807470440864563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.520705257775262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.9249656677246, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92258605957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13300, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29989221692085266, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.015749595873057843, + "grad_norm": 0.13058824837207794, + "kept_token_count": 572.7296875, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9801537156105041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.752475357847288e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.26929779052735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.24908447265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13310, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.29989258348941805, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.0189044626429677, + "grad_norm": 0.17077729105949402, + "kept_token_count": 573.0578125, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9801101565361023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.769487128825859e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.62183380126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.42225646972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13320, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.30056422352790835, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.01661251252517104, + "grad_norm": 0.08552706241607666, + "kept_token_count": 572.3421875, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9794095456600189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.043155539780856e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.43811798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3392639160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13330, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30144181549549104, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.014912061486393213, + "grad_norm": 0.12197510898113251, + "kept_token_count": 571.5359375, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9805678725242615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.590691166114993e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.45631408691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8313812255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13340, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.29990818798542024, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.014923922531306744, + "grad_norm": 0.12210570275783539, + "kept_token_count": 572.546875, + "learning_rate": 1.892379503427575e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9808590054512024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.476969549315981e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10414047241211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.70867919921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13350, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.30046434700489044, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.017759560979902743, + "grad_norm": 0.1270732581615448, + "kept_token_count": 572.4953125, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9807614266872406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.515087345382199e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.64439315795899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.41779174804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13360, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2983142852783203, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.012936619902029633, + "grad_norm": 0.08976161479949951, + "kept_token_count": 574.16875, + "learning_rate": 1.889186760191666e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9816625356674195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.16308320988901e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08443756103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.513134765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2999226897954941, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.014905410818755627, + "grad_norm": 0.11740625649690628, + "kept_token_count": 573.19375, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9808107733726501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.495804456993937e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.81385726928711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9490936279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13380, + "valid_token_count": 818.7625 + }, + { + "actual_token_mask_ratio": 0.2996483027935028, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.014916761964559554, + "grad_norm": 0.1595187485218048, + "kept_token_count": 573.209375, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9812885403633118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.309177381102927e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.78249664306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.48804931640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.30010915398597715, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.0160953932441771, + "grad_norm": 0.19288989901542664, + "kept_token_count": 572.6734375, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9804756462574005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.626721344422549e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.25035934448242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1709320068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13400, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.29936922788619996, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.01719590965658426, + "grad_norm": 0.12918226420879364, + "kept_token_count": 573.4171875, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9799008846282959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.851232658140361e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.63810806274414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7887725830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13410, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29951576590538026, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.017333672754466532, + "grad_norm": 0.11216998845338821, + "kept_token_count": 573.284375, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9809220612049103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.45234385249205e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38785934448242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0668426513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13420, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3005205154418945, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.017105771508067847, + "grad_norm": 0.1525431126356125, + "kept_token_count": 572.5828125, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9817943751811982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.11159773345571e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.63705291748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.37656555175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13430, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.2995729446411133, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.0166204777546227, + "grad_norm": 0.11562927812337875, + "kept_token_count": 573.1625, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9809516251087189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.44078919524327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.4457015991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5538360595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13440, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.2998658210039139, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.01496534449979663, + "grad_norm": 0.1652604639530182, + "kept_token_count": 572.93125, + "learning_rate": 1.876397026626085e-05, + "loss": 0.015, + "mgd_cos_sim": 0.980944675207138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.443505965056829e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.31368942260742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.337158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13450, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2999576270580292, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.014346807729452848, + "grad_norm": 0.10505956411361694, + "kept_token_count": 572.6984375, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9816483557224274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.168627344071865e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.78174362182617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.54188537597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13460, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29909059703350066, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.016202161181718112, + "grad_norm": 0.12878045439720154, + "kept_token_count": 573.4953125, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9800339281558991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.799258819431997e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.83372344970704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.14100341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13470, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3002980947494507, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.015702765341848134, + "grad_norm": 0.08152040094137192, + "kept_token_count": 572.6046875, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9810433149337768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.404975040117278e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.118994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0870635986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29919859766960144, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.014240528456866741, + "grad_norm": 0.08697819709777832, + "kept_token_count": 573.36875, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9804486334323883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.637265225639567e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.52621765136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66846923828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.3007379651069641, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.016135497204959392, + "grad_norm": 0.09753233939409256, + "kept_token_count": 572.084375, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9804618179798126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.63212599849794e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.28504943847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.22998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13500, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3007906496524811, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.01768604665994644, + "grad_norm": 0.09842593967914581, + "kept_token_count": 572.0140625, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9803116381168365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.690788625041023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.63181610107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2198883056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13510, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.299674054980278, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.013541866280138492, + "grad_norm": 0.13852821290493011, + "kept_token_count": 573.0046875, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.981237816810608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.32900392904412e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.77958374023437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.19078369140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13520, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.3007078468799591, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.014400982204824686, + "grad_norm": 0.1106601431965828, + "kept_token_count": 572.2140625, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.980403047800064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.655076187802478e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.54128036499023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5706817626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2998286455869675, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.015412142965942621, + "grad_norm": 0.11926517635583878, + "kept_token_count": 572.703125, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9821958243846893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.954771270102355e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.97530212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.74169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13540, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.30037400126457214, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.01617255937308073, + "grad_norm": 0.11531317979097366, + "kept_token_count": 572.534375, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9815856277942657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.193130695668515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.61352310180663, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.776171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13550, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2999803364276886, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.017086451686918736, + "grad_norm": 0.10586743801832199, + "kept_token_count": 572.8640625, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9798751533031463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.861293124733493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.8068229675293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.67535705566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13560, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3010870158672333, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.016612480394542217, + "grad_norm": 0.08419647067785263, + "kept_token_count": 571.946875, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.981065285205841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.396389337372966e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.10559768676758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7144348144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13570, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2993558496236801, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.016391081735491753, + "grad_norm": 0.10032570362091064, + "kept_token_count": 573.3546875, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9805489599704742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.598090087412857e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.26108779907227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5883544921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13580, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.30018588304519656, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.017765696719288827, + "grad_norm": 0.11378489434719086, + "kept_token_count": 572.4796875, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.980323177576065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.686276876484044e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.76008377075195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.74607543945314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13590, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.2989065945148468, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.01540324343368411, + "grad_norm": 0.15517202019691467, + "kept_token_count": 573.5609375, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9803995072841645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.65646058425773e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39967346191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.33277893066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13600, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30019234120845795, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.01334982244297862, + "grad_norm": 0.12796056270599365, + "kept_token_count": 572.7578125, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.979617565870285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.96190528490115e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.99217071533204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1422119140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13610, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29991187155246735, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.015683649480342864, + "grad_norm": 0.11453115195035934, + "kept_token_count": 572.725, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9800119459629059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.807852744008414e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.29040222167968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2167572021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13620, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.2996000647544861, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014656594581902027, + "grad_norm": 0.12437362223863602, + "kept_token_count": 573.321875, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.981319522857666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.297083793673664e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.43576431274414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2250946044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13630, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.29974921345710753, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.0138944401871413, + "grad_norm": 0.11837799102067947, + "kept_token_count": 573.1078125, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.981348204612732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.285877654794604e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.16789321899414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3549865722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13640, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.30072297155857086, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.01663803877308965, + "grad_norm": 0.13642294704914093, + "kept_token_count": 572.271875, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9802380383014679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.719537534285336e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.61852111816407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6612609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13650, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2998137205839157, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.014112479239702224, + "grad_norm": 0.1089320033788681, + "kept_token_count": 572.9875, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9818234503269195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.100233269738965e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.88558578491211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6960021972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.29887874722480773, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.015749753545969726, + "grad_norm": 0.11235272139310837, + "kept_token_count": 573.9390625, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9804118752479554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.651629493921064e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44309844970704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.60572509765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13670, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.3004304379224777, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.019093235861510037, + "grad_norm": 0.1788076013326645, + "kept_token_count": 572.325, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9804929912090301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.619945536134765e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.69536743164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.43641967773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2997924298048019, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.01631459854543209, + "grad_norm": 0.09892867505550385, + "kept_token_count": 572.971875, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9788957715034485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.243853953899815e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.52327194213868, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48839416503904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13690, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3004689961671829, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.016675826534628867, + "grad_norm": 0.10368466377258301, + "kept_token_count": 572.3171875, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9805906295776368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.581806712551043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.13664016723632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.55169677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13700, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.30000407695770265, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.016553069930523634, + "grad_norm": 0.13881643116474152, + "kept_token_count": 572.6484375, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9803054869174957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.693192892475054e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.98391189575196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0961486816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.2998075246810913, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.01408924963325262, + "grad_norm": 0.10357251763343811, + "kept_token_count": 573.203125, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9800738573074341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.78366629674565e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.9628288269043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.33331604003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.30046271085739135, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.018220947589725256, + "grad_norm": 0.11222042888402939, + "kept_token_count": 572.16875, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.980387282371521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.661241324967705e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.41925506591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.71129150390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13730, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.3002845346927643, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.013992624077945947, + "grad_norm": 0.10377689450979233, + "kept_token_count": 572.7859375, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9825146436691284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.830232123320457e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.53604202270508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9129943847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13740, + "valid_token_count": 818.59375 + }, + { + "actual_token_mask_ratio": 0.3002565324306488, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.01855610217899084, + "grad_norm": 0.1367844194173813, + "kept_token_count": 572.646875, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9813938498497009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.268046974786557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.11146392822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9082916259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13750, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3007340312004089, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.01677657039836049, + "grad_norm": 0.10841599851846695, + "kept_token_count": 572.1546875, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9798345685005188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.87713477620855e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.7802848815918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.21677856445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2997938275337219, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.013273802399635316, + "grad_norm": 0.11337630450725555, + "kept_token_count": 573.3140625, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.981321519613266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.296297953871545e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.1139762878418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.83192138671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13770, + "valid_token_count": 818.7765625 + }, + { + "actual_token_mask_ratio": 0.3000303119421005, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.01361895902082324, + "grad_norm": 0.11160998791456223, + "kept_token_count": 572.834375, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9812071800231934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.340966476476751e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.80384674072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9948364257813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2996834069490433, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.014951818715780973, + "grad_norm": 0.08437308669090271, + "kept_token_count": 572.8625, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9811724364757538, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.354530025622807e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.13431396484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.39296264648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13790, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2988452672958374, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.013989147171378136, + "grad_norm": 0.0892692506313324, + "kept_token_count": 573.6421875, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9809629857540131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.436348387273029e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.5949836730957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3115295410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2992389380931854, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.014394641388207674, + "grad_norm": 0.09967871010303497, + "kept_token_count": 573.571875, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9807713150978088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.51122282963479e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.57809753417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3185791015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13810, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29992270171642305, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.01646286714822054, + "grad_norm": 0.11202660202980042, + "kept_token_count": 572.7515625, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9793668270111084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.059847896220162e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.36408462524415, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.01507568359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13820, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.3000406801700592, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.01737045180052519, + "grad_norm": 0.11436553299427032, + "kept_token_count": 572.8359375, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9804810523986817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.62460389523767e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.50047988891602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4297729492188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13830, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29978591203689575, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.016932369396090507, + "grad_norm": 0.10375110059976578, + "kept_token_count": 573.0828125, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.017, + "mgd_cos_sim": 0.980544525384903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.599805248901249e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.25223541259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2213562011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13840, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2995584487915039, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015114446263760328, + "grad_norm": 0.1222154051065445, + "kept_token_count": 573.003125, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9809557974338532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.4391550879227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5312629699707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9974884033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13850, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.30012365281581876, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.012973517086356879, + "grad_norm": 0.11655212938785553, + "kept_token_count": 572.8453125, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9817333400249482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.135434716474265e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.75286254882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9423431396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13860, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.30028277039527895, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.01641125129535794, + "grad_norm": 0.0934268981218338, + "kept_token_count": 572.6796875, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9801334261894226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.760394291835837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.06892013549805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.974072265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13870, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.299809330701828, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.015982400812208652, + "grad_norm": 0.09179054945707321, + "kept_token_count": 573.1015625, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9814275681972504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.254876545630396e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.44853897094727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92810974121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13880, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.3005446881055832, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.015890573617070913, + "grad_norm": 0.10962659865617752, + "kept_token_count": 572.0984375, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.980817687511444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.493110606446863e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2417709350586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8997528076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13890, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.29937795996665956, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.014854706544429063, + "grad_norm": 0.09408421814441681, + "kept_token_count": 573.2765625, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.980188250541687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.738978456472978e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.43368072509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.030517578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13900, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.2994386196136475, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.01499854726716876, + "grad_norm": 0.1518884003162384, + "kept_token_count": 573.1203125, + "learning_rate": 1.802318716463359e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9815743744373322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.197527884272859e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.81200485229492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1133972167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.30126226544380186, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.014022504538297653, + "grad_norm": 0.10651349276304245, + "kept_token_count": 571.6203125, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9807060718536377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.536709963460453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.37728958129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.24493103027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13920, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3000111699104309, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.013570671901106834, + "grad_norm": 0.13015446066856384, + "kept_token_count": 572.615625, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9813204228878021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.296732946997509e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.68276824951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6512847900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13930, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.29928906857967374, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.013726377859711647, + "grad_norm": 0.10603737831115723, + "kept_token_count": 573.5515625, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9800533592700958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.7916761074448e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.58650131225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.93055114746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13940, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29941981434822085, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.01589419199153781, + "grad_norm": 0.08749940246343613, + "kept_token_count": 573.159375, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9812224268913269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.335009941016324e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.50403442382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0505004882813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13950, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.30010099411010743, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.015591076482087374, + "grad_norm": 0.10929425805807114, + "kept_token_count": 572.79375, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.981753820180893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.127424396458082e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.8700942993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5869903564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13960, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.30021404325962064, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.017248059064149855, + "grad_norm": 0.09334665536880493, + "kept_token_count": 572.4171875, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9820894539356232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.996323209023103e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.02214126586914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1508331298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13970, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.29951352477073667, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.013858051784336568, + "grad_norm": 0.1000179797410965, + "kept_token_count": 572.9171875, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9818783402442932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.07879225956276e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.12874221801758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.62940673828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13980, + "valid_token_count": 817.8828125 + }, + { + "actual_token_mask_ratio": 0.3008503168821335, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.014723994862288236, + "grad_norm": 0.10578280687332153, + "kept_token_count": 571.9078125, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9808555603027344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.478315965272486e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.56461029052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8979858398437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13990, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.301482355594635, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.012986478954553604, + "grad_norm": 0.0872153788805008, + "kept_token_count": 571.659375, + "learning_rate": 1.787728226180341e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9811173737049103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.376046341960319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.49684600830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.75738220214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14000, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.30003797113895414, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.015494519658386708, + "grad_norm": 0.13258160650730133, + "kept_token_count": 572.365625, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9806500196456909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.558598590549082e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.61563644409179, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4867828369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14010, + "valid_token_count": 817.7140625 + }, + { + "actual_token_mask_ratio": 0.29984393119812014, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.016305419150739908, + "grad_norm": 0.08453117311000824, + "kept_token_count": 572.9234375, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9809694051742553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.433844002662227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47714538574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0248718261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14020, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2998395025730133, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.013799062604084611, + "grad_norm": 0.06415016204118729, + "kept_token_count": 572.81875, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.981428998708725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.254323572851718e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.21620635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1708160400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14030, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30044379234313967, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.012822742015123368, + "grad_norm": 0.10140523314476013, + "kept_token_count": 572.4171875, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9812843441963196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.310820510610938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.0330421447754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.68418884277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14040, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30042438507080077, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.017856693267822264, + "grad_norm": 0.11010096222162247, + "kept_token_count": 572.5390625, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9809189438819885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.453560428984928e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.6877555847168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.24920959472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14050, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.30048092305660246, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.015415798965841531, + "grad_norm": 0.10608689486980438, + "kept_token_count": 572.290625, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9797996699810028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.890769265941344e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.18976821899415, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.87509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14060, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30037765204906464, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.012437191884964704, + "grad_norm": 0.20118869841098785, + "kept_token_count": 572.88125, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.980623745918274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.568866640212945e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.58363265991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.06105346679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14070, + "valid_token_count": 818.8453125 + }, + { + "actual_token_mask_ratio": 0.2994635015726089, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.01391154583543539, + "grad_norm": 0.10944539308547974, + "kept_token_count": 573.14375, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.014, + "mgd_cos_sim": 0.979742842912674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.91297308751382e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.30592575073243, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0334106445313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14080, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3001344293355942, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.014425180852413177, + "grad_norm": 0.09438174217939377, + "kept_token_count": 572.4046875, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9797612309455872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.905791499069892e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.15337448120117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9332763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14090, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.2993751734495163, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.014318202622234821, + "grad_norm": 0.10959506779909134, + "kept_token_count": 573.346875, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9807263970375061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.528768546762876e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82895355224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92225646972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14100, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.300131157040596, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.014692784380167722, + "grad_norm": 0.13888713717460632, + "kept_token_count": 572.4125, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9818369507789612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.094964021234774e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.46896133422851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0212341308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14110, + "valid_token_count": 817.8859375 + }, + { + "actual_token_mask_ratio": 0.29892506301403043, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.015911197196692227, + "grad_norm": 0.09233786165714264, + "kept_token_count": 573.71875, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9803500473499298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.67578967497684e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.57033843994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.03147277832034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14120, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2992202877998352, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.014989066123962402, + "grad_norm": 0.07960104197263718, + "kept_token_count": 573.409375, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9805086493492127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.613820489495992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.43869476318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.82789916992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14130, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3005932092666626, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.01366417109966278, + "grad_norm": 0.09757794439792633, + "kept_token_count": 572.3296875, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9819356441497803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.056401518639177e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.18244323730468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.81378173828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14140, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.30056374669075014, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.012554416432976723, + "grad_norm": 0.11008919030427933, + "kept_token_count": 572.5828125, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9813564896583558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.282635051524267e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99007263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4009735107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14150, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.29967956244945526, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.015341439004987478, + "grad_norm": 0.11319196969270706, + "kept_token_count": 573.0125, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9800992250442505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.773756515234709e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.020751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2864532470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14160, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.299125674366951, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.01498544504866004, + "grad_norm": 0.11066832393407822, + "kept_token_count": 573.55625, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9815487802028656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.207526432466693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01587219238282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4731750488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3006756126880646, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.01489663664251566, + "grad_norm": 0.11784189194440842, + "kept_token_count": 572.2484375, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9805395901203156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.60174123570323e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.91745529174804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.664404296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14180, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30059391260147095, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.01589190596714616, + "grad_norm": 0.12064152956008911, + "kept_token_count": 572.2734375, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9805334329605102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.604148486279882e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.0861213684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7264434814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.3007154673337936, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.015463493019342422, + "grad_norm": 0.09399595856666565, + "kept_token_count": 572.1484375, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9812410473823547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.327734492719173e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.91980285644532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0986785888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3010466009378433, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.015979920234531163, + "grad_norm": 0.14349953830242157, + "kept_token_count": 572.05, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.016, + "mgd_cos_sim": 0.980988472700119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.426395823131315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.03298950195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8917572021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14210, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2997220456600189, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.017127295769751073, + "grad_norm": 0.08608128875494003, + "kept_token_count": 572.8328125, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9809229075908661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.452004865626805e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.55945129394532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3146514892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14220, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.30072118937969206, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.01602257015183568, + "grad_norm": 0.14569337666034698, + "kept_token_count": 572.2640625, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9813674807548523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.27834580175113e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.89709854125977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4863647460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14230, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.29931764900684354, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.014552927017211914, + "grad_norm": 0.12217064946889877, + "kept_token_count": 573.1, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9811671316623688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.356610803981312e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.48540496826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0761779785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14240, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.30014878511428833, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.01630312604829669, + "grad_norm": 0.11399175226688385, + "kept_token_count": 572.4765625, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9813272058963776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.294075185200199e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.77179565429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2038879394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.3005766421556473, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.014463826268911361, + "grad_norm": 0.07917259633541107, + "kept_token_count": 572.3984375, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9809061706066131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.458547624992207e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07542114257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5614349365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14260, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3002455919981003, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.016189276706427336, + "grad_norm": 0.13863861560821533, + "kept_token_count": 572.5296875, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9812056362628937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.341573727899231e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.55256652832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39231567382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14270, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2993399441242218, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.014637702889740468, + "grad_norm": 0.11143612116575241, + "kept_token_count": 573.5046875, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9794618844985962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.02271744760219e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5023582458496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.58436889648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14280, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.3014284402132034, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.013919455232098699, + "grad_norm": 0.08540733903646469, + "kept_token_count": 571.671875, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9818075358867645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.10645821527578e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.21302490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.11582641601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3000989526510239, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.016250278800725937, + "grad_norm": 0.08654710650444031, + "kept_token_count": 572.703125, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9811655223369599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.3572329711169e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.1502082824707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7027526855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14300, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30051508247852327, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.018136603850871326, + "grad_norm": 0.08253923803567886, + "kept_token_count": 572.26875, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9809271037578583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.450363045791164e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.92859802246093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7460876464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14310, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30050782263278963, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.015406781248748303, + "grad_norm": 0.1208781972527504, + "kept_token_count": 572.5203125, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9817784428596497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.117816712707282e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.01625137329101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9292755126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14320, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2985380321741104, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.013645987305790187, + "grad_norm": 0.11551866680383682, + "kept_token_count": 574.0203125, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9792988240718842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.086415618890896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.46817550659179, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.87017822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14330, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3006534963846207, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.01806902913376689, + "grad_norm": 0.08301663398742676, + "kept_token_count": 571.9921875, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9810583472251893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.399100722977891e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.30530166625977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8325897216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14340, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.3001121371984482, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.012856408534571529, + "grad_norm": 0.08720581233501434, + "kept_token_count": 572.6609375, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9809711456298829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.43316879379563e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.7227294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4755096435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.299539515376091, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.014733912236988545, + "grad_norm": 0.0861562192440033, + "kept_token_count": 573.0953125, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9806148171424866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.572352478746325e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.9408950805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9901489257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14360, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29976987838745117, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.01334551041945815, + "grad_norm": 0.08140530437231064, + "kept_token_count": 573.2578125, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9810453295707703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.404190182569437e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.88514251708985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6446044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14370, + "valid_token_count": 818.6671875 + }, + { + "actual_token_mask_ratio": 0.3005550891160965, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.014835741836577654, + "grad_norm": 0.09604950249195099, + "kept_token_count": 572.546875, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9816338062286377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.1743151784176e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.41609420776368, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9899505615234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14380, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3005236029624939, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.01604994684457779, + "grad_norm": 0.08801977336406708, + "kept_token_count": 572.2203125, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9800629019737244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.787942886352539e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.63116226196288, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.38255004882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14390, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.2996312081813812, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.013874584436416626, + "grad_norm": 0.0997251346707344, + "kept_token_count": 573.046875, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9812372148036956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.329228610615246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1146369934082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4864166259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14400, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2998222321271896, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.01415753122419119, + "grad_norm": 0.09918683767318726, + "kept_token_count": 572.85, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9810291111469269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.410519901895896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.18101806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3548309326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14410, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.30022871792316436, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.013196052564308048, + "grad_norm": 0.07207247614860535, + "kept_token_count": 572.471875, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9810066521167755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.419298417516984e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.29437866210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.28118286132815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14420, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3005896985530853, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.014726046659052371, + "grad_norm": 0.08620718121528625, + "kept_token_count": 572.1296875, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9794896483421326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 8.011874670046381e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.54392013549804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0516418457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14430, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.30010342597961426, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.01620312137529254, + "grad_norm": 0.12582866847515106, + "kept_token_count": 572.68125, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9823065102100372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.91154222295154e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21835174560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7370666503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14440, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3007338374853134, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.013291920628398656, + "grad_norm": 0.10154587030410767, + "kept_token_count": 572.2703125, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9818013787269593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.108854842954315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.24411544799804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1578826904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14450, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2998352497816086, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.013855133205652237, + "grad_norm": 0.10607337206602097, + "kept_token_count": 572.63125, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9819703698158264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.04284346284112e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.22399673461913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4351104736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14460, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.2981801688671112, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.013968027010560036, + "grad_norm": 0.0974210649728775, + "kept_token_count": 574.44375, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9807791590690613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.508163616876118e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.91572799682618, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.23009338378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14470, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2994658946990967, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.0127465826459229, + "grad_norm": 0.10442717373371124, + "kept_token_count": 573.434375, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9816083490848542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.184260612120852e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.2214256286621, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9233917236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14480, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.3003057986497879, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.014658936765044928, + "grad_norm": 0.07064204663038254, + "kept_token_count": 572.4828125, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9806928038597107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.541888262494468e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.656005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8261322021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14490, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2997052013874054, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.013781946245580911, + "grad_norm": 0.163430318236351, + "kept_token_count": 573.1765625, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9814117312431335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.261056452989578e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.5575439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6266571044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14500, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.2994188219308853, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.01665136041119695, + "grad_norm": 0.1457502245903015, + "kept_token_count": 573.278125, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9804933726787567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.619789794262033e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.20890731811524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4290252685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14510, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2989898532629013, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.014712680131196976, + "grad_norm": 0.11569222807884216, + "kept_token_count": 573.784375, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9809259474277496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.45082288631238e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61529388427735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1166229248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14520, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.29973383247852325, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.014779252745211125, + "grad_norm": 0.09206170588731766, + "kept_token_count": 572.6796875, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9811964511871338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.345154590439051e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.74855575561523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3142578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14530, + "valid_token_count": 817.80625 + }, + { + "actual_token_mask_ratio": 0.30022192001342773, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.013789546117186547, + "grad_norm": 0.11316948384046555, + "kept_token_count": 572.63125, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9813607633113861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.28097598766908e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.41330184936524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9151184082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14540, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2996846854686737, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.015536900795996188, + "grad_norm": 0.1011853814125061, + "kept_token_count": 573.0359375, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9811236262321472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.373606640612707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.21506576538086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9246398925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14550, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2996888428926468, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.015497935563325882, + "grad_norm": 0.08947280049324036, + "kept_token_count": 572.940625, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9808936655521393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.463428773917258e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.93161544799804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.980615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14560, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30049424171447753, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.014290561992675066, + "grad_norm": 0.06947632879018784, + "kept_token_count": 572.3703125, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9815246522426605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.216951416921802e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.22174072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.11488342285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14570, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29981144666671755, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.014127112319692969, + "grad_norm": 0.09347674250602722, + "kept_token_count": 572.91875, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9808327317237854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.487235052394681e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.40542144775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.89405212402346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14580, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30051096975803376, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.014566301461309194, + "grad_norm": 0.15000177919864655, + "kept_token_count": 572.3546875, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9807627677917481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.514559692936018e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.16042938232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7983856201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14590, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3010831087827682, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.013287666626274586, + "grad_norm": 0.0819631963968277, + "kept_token_count": 571.878125, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9823558092117309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.892279125167989e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.61352767944337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8134399414063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14600, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29886373579502107, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.01493886774405837, + "grad_norm": 0.10812769830226898, + "kept_token_count": 573.5296875, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9808206379413604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.491959477192722e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3053192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14610, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.29975988864898684, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.01370596531778574, + "grad_norm": 0.12649095058441162, + "kept_token_count": 572.859375, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9810059368610382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.419580069836229e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.98802185058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.24098510742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14620, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.3003368675708771, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.013724375143647194, + "grad_norm": 0.0874936431646347, + "kept_token_count": 572.55, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9811831653118134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.350347004830837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.96964721679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.8273864746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.30016996562480924, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.014546554815024138, + "grad_norm": 0.10436279326677322, + "kept_token_count": 572.603125, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9814776837825775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.235303055495024e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.47115631103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.46116638183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14640, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2999161660671234, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.01399251027032733, + "grad_norm": 0.1189674511551857, + "kept_token_count": 572.925, + "learning_rate": 1.681600449102196e-05, + "loss": 0.014, + "mgd_cos_sim": 0.979602974653244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.967604979057797e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.23382797241212, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6206481933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2998171538114548, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.0143224167637527, + "grad_norm": 0.1360686868429184, + "kept_token_count": 572.978125, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9811637043952942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.357941940426827e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.90642700195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.57581787109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14660, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29908588230609895, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.017726767156273127, + "grad_norm": 0.10125011205673218, + "kept_token_count": 573.6296875, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9804754018783569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.626807928318158e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.79852600097657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9374053955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14670, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2993606597185135, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.015529511868953705, + "grad_norm": 0.10724987089633942, + "kept_token_count": 573.5125, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.980349338054657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.676057939534075e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.04368743896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5339050292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14680, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.2994419127702713, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.01351219303905964, + "grad_norm": 0.09536203742027283, + "kept_token_count": 573.203125, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9810505211353302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.40216339181643e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.42211227416992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8287750244141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14690, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29910852015018463, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.016792534198611975, + "grad_norm": 0.11785581707954407, + "kept_token_count": 573.621875, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9810460925102233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.403882482321933e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.9925048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4045349121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14700, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.30001831650733946, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.013403603434562683, + "grad_norm": 0.09034402668476105, + "kept_token_count": 572.93125, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9808126628398895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.495069148717448e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.44825057983398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7658233642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14710, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.30059848427772523, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.015234849881380796, + "grad_norm": 0.11050038784742355, + "kept_token_count": 572.6234375, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9798126101493836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.885716404416598e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.8353271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4178070068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14720, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.30018773674964905, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.016906552482396364, + "grad_norm": 0.12542018294334412, + "kept_token_count": 572.709375, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9812437653541565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.326675113290548e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.4733283996582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8371887207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14730, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30071812570095063, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.013977947272360325, + "grad_norm": 0.0919579342007637, + "kept_token_count": 572.0109375, + "learning_rate": 1.666819987555675e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9809455513954163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.443155118380673e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.97410888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0147277832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14740, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.3001632660627365, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.013209903426468373, + "grad_norm": 0.10383383184671402, + "kept_token_count": 572.50625, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9815755665302277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.19706091331318e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.96715545654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62122802734376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14750, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.30024164021015165, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.015432203933596612, + "grad_norm": 0.10723475366830826, + "kept_token_count": 572.9109375, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9818897664546966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.074331151670776e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.81237182617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.63751220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14760, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.30097473561763766, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.01421710168942809, + "grad_norm": 0.09492649137973785, + "kept_token_count": 572.29375, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9815966308116912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.188834497355856e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.80998001098632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.03810729980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14770, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.3007712662220001, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.01426738416776061, + "grad_norm": 0.10871335119009018, + "kept_token_count": 572.109375, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9810838103294373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.389154779957607e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.74356994628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.033154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14780, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.3010705322027206, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.013945532031357288, + "grad_norm": 0.12320752441883087, + "kept_token_count": 571.909375, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.014, + "mgd_cos_sim": 0.980622160434723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.569485533167608e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 78.3165397644043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.51202392578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14790, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.30023784637451173, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.015392884705215692, + "grad_norm": 0.08689864724874496, + "kept_token_count": 572.49375, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9813404619693756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.288895249075722e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23461685180663, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.58384094238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14800, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3001603037118912, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.015133301820605993, + "grad_norm": 0.0931585282087326, + "kept_token_count": 572.8875, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9806112468242645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.573742768727243e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.37174682617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9473175048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14810, + "valid_token_count": 818.5984375 + }, + { + "actual_token_mask_ratio": 0.30045532882213594, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.012508462369441985, + "grad_norm": 0.07945437729358673, + "kept_token_count": 572.571875, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9815731883049011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.197986633400433e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3142120361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14820, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.30073406398296354, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.0160601825453341, + "grad_norm": 0.09089966118335724, + "kept_token_count": 572.2921875, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9798517405986786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.870436747907661e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.1825439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3437530517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14830, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2993137717247009, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.01506366180256009, + "grad_norm": 0.10680699348449707, + "kept_token_count": 573.1921875, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9817089855670929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.144940609578043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.9337257385254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.52214965820315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14840, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.2997874528169632, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.012533203139901162, + "grad_norm": 0.08433791995048523, + "kept_token_count": 572.9546875, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9807231068611145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.530054717790336e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.46372451782227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9978973388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14850, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30036700963974, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.01510692061856389, + "grad_norm": 0.1141672357916832, + "kept_token_count": 572.5828125, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9817708671092987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.120767986634746e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.40517807006836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5160369873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14860, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3006304293870926, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.01497382139787078, + "grad_norm": 0.10701168328523636, + "kept_token_count": 572.3046875, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9803951263427735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.658174363314174e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.62548217773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7294616699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14870, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.3006743282079697, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.016140959784388544, + "grad_norm": 0.1640230268239975, + "kept_token_count": 572.2046875, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.981161504983902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.358802904491312e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.39578094482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3116912841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14880, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3002876043319702, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.014319628663361072, + "grad_norm": 0.09509419649839401, + "kept_token_count": 572.715625, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9810947299003601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.384888449450955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9534080505371, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.89232177734374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14890, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.29975695013999937, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.017213101964443922, + "grad_norm": 0.13481079041957855, + "kept_token_count": 573.1578125, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9813678503036499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.278209959622472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.42237396240235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.71766052246096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14900, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.29993989765644075, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.01601310893893242, + "grad_norm": 0.10299088805913925, + "kept_token_count": 572.7296875, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9806813597679138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.546361011918634e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.89005050659179, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2473419189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2999744683504105, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.013992493133991956, + "grad_norm": 0.10090067982673645, + "kept_token_count": 572.8, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9818083703517914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.10612461261917e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48830032348633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.62851867675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14920, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3002903938293457, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.017309982050210237, + "grad_norm": 0.14932939410209656, + "kept_token_count": 572.55625, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9803259432315826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.685197924729436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.51357345581054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.58251037597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14930, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.299527582526207, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.0156942093744874, + "grad_norm": 0.09634274244308472, + "kept_token_count": 573.2609375, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.980921882390976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.45241210097447e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.18805313110352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.87949523925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14940, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2999296814203262, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.014412773866206407, + "grad_norm": 0.11974233388900757, + "kept_token_count": 572.8625, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9812748014926911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.314551112358458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.4385871887207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9010467529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14950, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.3009100705385208, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.013096281653270125, + "grad_norm": 0.08953956514596939, + "kept_token_count": 572.228125, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9811721324920655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.354647023021244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.62494888305665, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8655151367187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14960, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.299432760477066, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.014998219348490239, + "grad_norm": 0.09101346880197525, + "kept_token_count": 573.6, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9799707472324372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.823949563317001e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.49719619750977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.607373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14970, + "valid_token_count": 818.7609375 + }, + { + "actual_token_mask_ratio": 0.29932203590869905, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.0148497411981225, + "grad_norm": 0.07911534607410431, + "kept_token_count": 573.1671875, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9815780043601989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.196105761977378e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.44271697998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.86082458496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14980, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2986560523509979, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.014254711661487817, + "grad_norm": 0.10972604155540466, + "kept_token_count": 573.8453125, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9817759394645691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.118794455891475e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.59915466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9595489501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14990, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30085036158561707, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.014963129349052907, + "grad_norm": 0.13348355889320374, + "kept_token_count": 572.1734375, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.015, + "mgd_cos_sim": 0.981258225440979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.321016819332727e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.99364852905273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6405487060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15000, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.30044602751731875, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.016533240489661694, + "grad_norm": 0.11903458833694458, + "kept_token_count": 572.5828125, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9811355710029602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.368937804130837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.06845245361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4338897705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15010, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29984337091445923, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.015526504535228015, + "grad_norm": 0.12435057014226913, + "kept_token_count": 573.003125, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9812900185585022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.308602798730134e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.80341110229492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.52255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15020, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.30053567588329316, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.013278234750032425, + "grad_norm": 0.12028281390666962, + "kept_token_count": 572.3859375, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.981042492389679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.405292271869257e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.78913345336915, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8033843994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15030, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.299970817565918, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.013294253684580325, + "grad_norm": 0.10965467244386673, + "kept_token_count": 572.875, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9810897827148437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.386823708657175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.99127578735352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.92261657714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.29935153424739835, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.014100223779678345, + "grad_norm": 0.10404621064662933, + "kept_token_count": 573.44375, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9818413138389588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.093260137480684e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.94749526977539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9864044189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15050, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.2995411902666092, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.014314059866592288, + "grad_norm": 0.08115222305059433, + "kept_token_count": 573.2875, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9825994670391083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.79710297845304e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.91972885131835, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4073455810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15060, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.3000000476837158, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.016078670788556338, + "grad_norm": 0.1223631426692009, + "kept_token_count": 572.6453125, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9816076278686523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.184544410847593e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36724472045898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5473327636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15070, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.3000532925128937, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.01734649958088994, + "grad_norm": 0.12680859863758087, + "kept_token_count": 572.8234375, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9811296701431275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.37124039005721e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6517463684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.58734130859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15080, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.3006899684667587, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.01680966466665268, + "grad_norm": 0.11787901818752289, + "kept_token_count": 572.1234375, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9813847124576569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.271619542734698e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2339599609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2965118408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15090, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.29912050664424894, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.015274533256888389, + "grad_norm": 0.10685572028160095, + "kept_token_count": 573.4078125, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9819869995117188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.036349743430037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.40599136352539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7647308349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15100, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.29953467547893525, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.016074651293456553, + "grad_norm": 0.09446930140256882, + "kept_token_count": 572.9890625, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9811595678329468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.359563969657756e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25734252929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.39292602539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15110, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2992706924676895, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.014811167772859336, + "grad_norm": 0.1401240974664688, + "kept_token_count": 573.5453125, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9825390160083771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.820719354436733e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.3760498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.53936462402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15120, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.3007060021162033, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.01289860149845481, + "grad_norm": 0.10631191730499268, + "kept_token_count": 572.4140625, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9812663793563843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.317838899325579e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7406997680664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6465301513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15130, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.2998866498470306, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.016275922674685715, + "grad_norm": 0.16329585015773773, + "kept_token_count": 573.040625, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9802061259746552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.732000231044366e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.90670013427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6878326416016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15140, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.29911175072193147, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.013363089319318533, + "grad_norm": 0.08853725343942642, + "kept_token_count": 573.334375, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9815289556980134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.21527037967462e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99622116088867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.15447387695315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15150, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.3001770079135895, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.013956966483965516, + "grad_norm": 0.1223938837647438, + "kept_token_count": 572.428125, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9800939917564392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.775802441756241e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.84195556640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.81341247558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15160, + "valid_token_count": 817.965625 + }, + { + "actual_token_mask_ratio": 0.2999032586812973, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.01564055485650897, + "grad_norm": 0.08584018796682358, + "kept_token_count": 572.74375, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9810033798217773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.420575566357002e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.78187255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3578796386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15170, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.29977419078350065, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.015006079617887736, + "grad_norm": 0.11104016751050949, + "kept_token_count": 573.1203125, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9805373311042785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.602627047162969e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.71190795898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.963037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15180, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2998787581920624, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.01733690155670047, + "grad_norm": 0.09416656196117401, + "kept_token_count": 572.8640625, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9807885169982911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.504505047108977e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72706909179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9943603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.30014537274837494, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.01524834642186761, + "grad_norm": 0.10181336104869843, + "kept_token_count": 572.5203125, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9820579349994659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.008636530372314e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.00383834838867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7645782470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15200, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2989520341157913, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.013956165593117475, + "grad_norm": 0.0944552943110466, + "kept_token_count": 573.7796875, + "learning_rate": 1.589399705694593e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9817556381225586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.126723212422804e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.70394439697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6034851074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15210, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.29973816573619844, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.014598180446773768, + "grad_norm": 0.0854387953877449, + "kept_token_count": 572.8109375, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9813321471214295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.292142472579144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.27296829223633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.31806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15220, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29955786764621734, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.01541251977905631, + "grad_norm": 0.10233771800994873, + "kept_token_count": 573.1578125, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9815033614635468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.225263834698126e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.72273025512695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3313842773438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15230, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.3004949301481247, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.015572550613433123, + "grad_norm": 0.14327488839626312, + "kept_token_count": 572.4640625, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9811729192733765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.354355839197523e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91393203735352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.77344970703126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.30022388994693755, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.014469103887677193, + "grad_norm": 0.1254781037569046, + "kept_token_count": 572.4296875, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9815822660923004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.194438912847546e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.19423904418946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6704833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15250, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.29940518736839294, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.013834774028509855, + "grad_norm": 0.10216648131608963, + "kept_token_count": 573.19375, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.98086998462677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.472681463696062e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.44936904907226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9588928222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15260, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.30017862021923064, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.017878071777522563, + "grad_norm": 0.0874113067984581, + "kept_token_count": 572.590625, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9802681982517243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.707753029535524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75365142822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.25641479492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15270, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2999581754207611, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.015260480716824532, + "grad_norm": 0.08370176702737808, + "kept_token_count": 572.903125, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9813201785087585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.296819676412269e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.00983047485352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.32627563476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15280, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3007281571626663, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.01422866778448224, + "grad_norm": 0.10551578551530838, + "kept_token_count": 572.153125, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9821643531322479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.967068948142696e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.31635894775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.87811889648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15290, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.2997405558824539, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.015508483862504363, + "grad_norm": 0.12208953499794006, + "kept_token_count": 573.0109375, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9804395854473114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.64080323278904e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6840805053711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.21686096191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15300, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30020143985748293, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.013888971926644445, + "grad_norm": 0.1038070023059845, + "kept_token_count": 572.6421875, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9807059407234192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.536764387623407e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.42211685180663, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3224670410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15310, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30001345574855803, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.012144386302679777, + "grad_norm": 0.07423914223909378, + "kept_token_count": 572.8421875, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9809367477893829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.446598865499255e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.40974578857421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4556121826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15320, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2993882864713669, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.01432631341740489, + "grad_norm": 0.10327291488647461, + "kept_token_count": 573.084375, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9816698670387268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.160229361033999e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.94345016479492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.10558166503904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15330, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.3001154690980911, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.012949493061751128, + "grad_norm": 0.10826334357261658, + "kept_token_count": 572.68125, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9823468744754791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.895768019603566e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.99183731079101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.20781860351565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15340, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29990221858024596, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.014668471459299327, + "grad_norm": 0.08487039059400558, + "kept_token_count": 572.7921875, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9821233153343201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.983095736359246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9639175415039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.32280883789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15350, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.29985177516937256, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.015172337181866169, + "grad_norm": 0.09527932852506638, + "kept_token_count": 573.1140625, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9821101725101471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.988231543800794e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.90926361083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1649230957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15360, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.30080742835998536, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.016372116282582283, + "grad_norm": 0.11349078267812729, + "kept_token_count": 572.2046875, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9813942849636078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.26787286112085e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.558349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5862731933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15370, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.29917085766792295, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.014291587192565203, + "grad_norm": 0.10105179250240326, + "kept_token_count": 573.3828125, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9816683173179627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.160830500652082e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.03373947143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.30609741210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15380, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29965207278728484, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.016091802157461643, + "grad_norm": 0.11714092642068863, + "kept_token_count": 573.05625, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9806940853595734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.541389932157472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.37472381591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1985229492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15390, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30069742500782015, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.014492228999733925, + "grad_norm": 0.10203582793474197, + "kept_token_count": 572.428125, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9815352559089661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.212806958705187e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.47306060791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8000915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15400, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.29973365664482116, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.012620506901293993, + "grad_norm": 0.09445300698280334, + "kept_token_count": 572.978125, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9810026228427887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.420865294989198e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.91969451904296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.89524841308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15410, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30060455203056335, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.01637344704940915, + "grad_norm": 0.1024329736828804, + "kept_token_count": 572.390625, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9809254169464111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.451027340721339e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.37950744628907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5181365966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15420, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2995735138654709, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.014011543057858945, + "grad_norm": 0.1059783548116684, + "kept_token_count": 573.2078125, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9807395219802857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.523642852902412e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17302932739258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.30267944335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15430, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.2989783376455307, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.016962672770023345, + "grad_norm": 0.10614671558141708, + "kept_token_count": 573.75, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9811927735805511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.346588608925231e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.67848510742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.38926391601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15440, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.3002736508846283, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.013734650611877442, + "grad_norm": 0.10100962221622467, + "kept_token_count": 572.6609375, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9808156192302704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.493917219107971e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.79012756347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8064239501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15450, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.299662384390831, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.014233287144452333, + "grad_norm": 0.09268993884325027, + "kept_token_count": 572.959375, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9808106780052185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.495845929952338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.35151672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5394256591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15460, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2996172785758972, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.012397472094744445, + "grad_norm": 0.08585377782583237, + "kept_token_count": 573.1609375, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9806298613548279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.56647867092397e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.32208938598633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3551422119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3005701839923859, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.015802962239831686, + "grad_norm": 0.09371532499790192, + "kept_token_count": 571.9296875, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9813391625881195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.289404311450198e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21709442138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.91314086914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15480, + "valid_token_count": 817.709375 + }, + { + "actual_token_mask_ratio": 0.2991244286298752, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.015039278008043767, + "grad_norm": 0.0820939689874649, + "kept_token_count": 573.40625, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.981012761592865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.416916414513253e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.95961990356446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.44334411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15490, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.299924299120903, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.01617421368137002, + "grad_norm": 0.08621539920568466, + "kept_token_count": 572.7484375, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9808353245258331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.486223184969277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.21974258422851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3456573486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15500, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.299905052781105, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.013742727972567082, + "grad_norm": 0.09393518418073654, + "kept_token_count": 573.0296875, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9815492451190948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.207343660411424e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.33473129272461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.86919860839845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15510, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.29960761666297914, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.015739587787538766, + "grad_norm": 0.11878390610218048, + "kept_token_count": 572.9890625, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9810825943946838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.389635866275057e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.80482025146485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6351043701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15520, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.299142187833786, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.012897336715832352, + "grad_norm": 0.1379023790359497, + "kept_token_count": 573.3984375, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9823073148727417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.911221680638846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.07106018066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5494079589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15530, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.3006513148546219, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.015043694619089365, + "grad_norm": 0.10064338892698288, + "kept_token_count": 571.95, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9803787171840668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.664586810278707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.30529174804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.635205078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15540, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.2992620557546616, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.01587412878870964, + "grad_norm": 0.08408370614051819, + "kept_token_count": 573.6296875, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9816023468971252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.1866038342705e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.20410690307617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0032897949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15550, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.29977172911167144, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.01585932159796357, + "grad_norm": 0.09714720398187637, + "kept_token_count": 572.840625, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9814557075500489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.243881045724266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.6420051574707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.99132385253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15560, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.3000242382287979, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.01683102622628212, + "grad_norm": 0.11204405874013901, + "kept_token_count": 572.7921875, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9812437176704407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.326695340452716e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.5568977355957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20252380371096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15570, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.300598081946373, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.015407008118927479, + "grad_norm": 0.07959004491567612, + "kept_token_count": 572.6890625, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9818598449230194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.086019541020505e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.40141372680664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.21490478515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15580, + "valid_token_count": 818.815625 + }, + { + "actual_token_mask_ratio": 0.3011515259742737, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.0139527240768075, + "grad_norm": 0.10988125950098038, + "kept_token_count": 571.89375, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9815406978130341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.210687908809632e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.88239059448242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0360595703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15590, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.30028128027915957, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.015760481264442204, + "grad_norm": 0.09582912176847458, + "kept_token_count": 572.8140625, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.980529373884201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.605734062963166e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.77062072753907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3367889404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15600, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.30001904964447024, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.014546331018209457, + "grad_norm": 0.13784612715244293, + "kept_token_count": 573.046875, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9823084592819213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.910779266036115e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.26555023193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4345275878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15610, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.29932118356227877, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.013173269294202327, + "grad_norm": 0.11736967414617538, + "kept_token_count": 573.36875, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9814403295516968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.249885820783674e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.85358505249023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3181549072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15620, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.3012382835149765, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.013225267641246319, + "grad_norm": 0.10826548933982849, + "kept_token_count": 571.8046875, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9816826999187469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.15522030077409e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.61367416381836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7576934814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15630, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.30083896219730377, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.014901068713515996, + "grad_norm": 0.10572002828121185, + "kept_token_count": 572.26875, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.980803805589676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.49852872104384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.66200561523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2130889892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15640, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.3000171631574631, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.013571907207369805, + "grad_norm": 0.10019746422767639, + "kept_token_count": 572.771875, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9811984598636627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.344366385950707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.28723907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9808288574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15650, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.29969883263111113, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.01542652426287532, + "grad_norm": 0.10781592130661011, + "kept_token_count": 573.2203125, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9816567778587342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.165338029153645e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.31369247436524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.74725952148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15660, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.30155143737792967, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.01305335881188512, + "grad_norm": 0.09777072817087173, + "kept_token_count": 571.703125, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9818122506141662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.104605101631023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.51104202270508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9010040283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15670, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.29968812465667727, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.01342418803833425, + "grad_norm": 0.10727852582931519, + "kept_token_count": 573.0140625, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.980837094783783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.485523747163825e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.18384628295898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7849639892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30029930770397184, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.012895892839878798, + "grad_norm": 0.10461241751909256, + "kept_token_count": 572.515625, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9814936101436615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.229083785205148e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.47684326171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3374969482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15690, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.30018719732761384, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.013870761636644602, + "grad_norm": 0.09636881947517395, + "kept_token_count": 572.5203125, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.980400276184082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.656164088984952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.24912033081054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0097991943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15700, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.301267147064209, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.01474828845821321, + "grad_norm": 0.11301634460687637, + "kept_token_count": 571.55, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.981261134147644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.319888973142952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36662368774414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.99143981933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15710, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.3003739476203918, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013446255680173635, + "grad_norm": 0.14068233966827393, + "kept_token_count": 572.534375, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9818193733692169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.101825249264949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.55810775756837, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.30311279296876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29921476542949677, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.014321292471140622, + "grad_norm": 0.10030058026313782, + "kept_token_count": 573.4828125, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.980869060754776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.473040823242627e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.36057510375977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.95118713378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15730, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2994280785322189, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.013645029114559292, + "grad_norm": 0.1363675445318222, + "kept_token_count": 573.134375, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9816965579986572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.149803532229271e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77522735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.007373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15740, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30071231424808503, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.015335951559245586, + "grad_norm": 0.09946059435606003, + "kept_token_count": 571.975, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9820651829242706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.005805018707178e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.82029190063477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0656677246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15750, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.3002340942621231, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.014730570744723082, + "grad_norm": 0.13025185465812683, + "kept_token_count": 572.553125, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9805400311946869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.601569159305654e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.32594375610351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4486907958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15760, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2998701184988022, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.015053954906761646, + "grad_norm": 0.09129302948713303, + "kept_token_count": 572.834375, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9824151635169983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.86909774231026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.33837814331055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.02703857421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.3002390444278717, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.015023926878347993, + "grad_norm": 0.1307811737060547, + "kept_token_count": 572.38125, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9817187547683716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.141133464756421e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96200332641601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5407409667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15780, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.2998800456523895, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.015322214737534524, + "grad_norm": 0.08933106809854507, + "kept_token_count": 572.86875, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9825822412967682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.803827418480068e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.08064575195313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9118377685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2999283760786057, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.015652642864733935, + "grad_norm": 0.11060334742069244, + "kept_token_count": 573.053125, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9822660744190216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.9273347617127e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.12849426269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0571350097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15800, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3004757821559906, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.01464396733790636, + "grad_norm": 0.09918110817670822, + "kept_token_count": 572.2671875, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9810615956783295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.397835652227514e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.1326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4730560302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15810, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.2991515815258026, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.014464741665869951, + "grad_norm": 0.14179493486881256, + "kept_token_count": 573.3828125, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9813275933265686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.293924791156315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61872253417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.64671936035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15820, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.2998274952173233, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.013410994410514831, + "grad_norm": 0.12080669403076172, + "kept_token_count": 572.9484375, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9817216157913208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.140008165151812e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.71160659790038, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4070129394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15830, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.299578058719635, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.014122941438108683, + "grad_norm": 0.11996522545814514, + "kept_token_count": 573.1453125, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9809784233570099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.430323894368485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.42756958007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.53388671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15840, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.30110354721546173, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.015244141593575478, + "grad_norm": 0.08809047192335129, + "kept_token_count": 571.840625, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9817427217960357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.131767088139896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.06754302978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0865417480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15850, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30039496123790743, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.017927466612309217, + "grad_norm": 0.10438297688961029, + "kept_token_count": 572.509375, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9812496066093445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.324388207052834e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.22788009643554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.30255432128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15860, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30074010491371156, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.01482592886313796, + "grad_norm": 0.08737670630216599, + "kept_token_count": 572.115625, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9820428371429444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.014536822680384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.32962341308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.13885803222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15870, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2997846961021423, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.015168740972876548, + "grad_norm": 0.09431320428848267, + "kept_token_count": 572.9765625, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9815201997756958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.218694663606584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.77963943481446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.68943481445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15880, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2995820134878159, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.016721028462052347, + "grad_norm": 0.12306943535804749, + "kept_token_count": 572.8703125, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9826375842094421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.78221636917442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.84456634521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7494720458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15890, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.3003128856420517, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.013555696886032819, + "grad_norm": 0.11009448021650314, + "kept_token_count": 572.425, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.981601881980896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.186780785559677e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.69205856323242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6851837158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15900, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.3000934809446335, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.013991471007466316, + "grad_norm": 0.12133440375328064, + "kept_token_count": 572.934375, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.014, + "mgd_cos_sim": 0.981622964143753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.178548985393718e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.12344894409179, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1225860595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15910, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.30027152597904205, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.014069446083158254, + "grad_norm": 0.12119419127702713, + "kept_token_count": 572.771875, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9814825654029846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.233392534544691e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93748779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4489410400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15920, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.3000650018453598, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.014931060187518597, + "grad_norm": 0.11576773226261139, + "kept_token_count": 572.9, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9802767515182496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.704407835262828e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.65815887451171, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8202880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15930, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.3008680909872055, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.013663744367659093, + "grad_norm": 0.11106003075838089, + "kept_token_count": 572.240625, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9825095117092133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.83223843225278e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.86902389526367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7100524902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15940, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.2999800264835358, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.01595152234658599, + "grad_norm": 0.11336740851402283, + "kept_token_count": 572.91875, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9818557858467102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.08760468114633e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.71921157836914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0823669433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15950, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30002833604812623, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.013454919774085283, + "grad_norm": 0.08320438116788864, + "kept_token_count": 572.915625, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9808235347270966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.49082159018144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.32116928100587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.79950256347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15960, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.30025765001773835, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.014184564910829067, + "grad_norm": 0.10209447890520096, + "kept_token_count": 572.7640625, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9818103313446045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.105362383299507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39919891357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.62205810546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15970, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.3002600848674774, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.015586248505860567, + "grad_norm": 0.138621523976326, + "kept_token_count": 572.5734375, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9807767689228057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.509087809012271e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.69645233154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.65892333984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15980, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.3009575068950653, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.014665581192821265, + "grad_norm": 0.12303055077791214, + "kept_token_count": 572.1078125, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9811700403690338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.355469460890163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.68778686523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8875701904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15990, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.3006693869829178, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.013349931128323079, + "grad_norm": 0.08594990521669388, + "kept_token_count": 572.4578125, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9820466935634613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.013028298388235e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.17184219360351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9525451660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16000, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.2996604979038239, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.012829771637916565, + "grad_norm": 0.11836843937635422, + "kept_token_count": 572.90625, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9801551699638367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.751906523481011e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.50369110107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.06017150878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16010, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.3007974445819855, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.015427533071488142, + "grad_norm": 0.13710053265094757, + "kept_token_count": 571.99375, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9806217670440673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.569641602458432e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.31512069702148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.25274658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16020, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29901584386825564, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.012598826829344035, + "grad_norm": 0.09902907907962799, + "kept_token_count": 573.7671875, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9815612733364105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.20265361451311e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.81301956176758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1605926513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16030, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.29993564188480376, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.014187768567353487, + "grad_norm": 0.09482783079147339, + "kept_token_count": 572.89375, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9815341472625733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.213243734440766e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08886184692383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.26487731933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16040, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2990288108587265, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.014834614284336568, + "grad_norm": 0.13099145889282227, + "kept_token_count": 573.4828125, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.981607049703598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.184765345300548e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.70910034179687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.50415954589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.3006717026233673, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.015732174552977086, + "grad_norm": 0.11276718974113464, + "kept_token_count": 572.0234375, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.981780868768692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.1168636350194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.27753829956055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9145904541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16060, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.30047090351581573, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.01375413197092712, + "grad_norm": 0.10237482935190201, + "kept_token_count": 572.51875, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9811197459697724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.375122368102893e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.33210067749023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6780029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.3013408899307251, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.015251992736011744, + "grad_norm": 0.0967848002910614, + "kept_token_count": 571.5359375, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9806721568107605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.549959482275881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.43926544189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.93885192871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16080, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.3002528190612793, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.01793446522206068, + "grad_norm": 0.09328778088092804, + "kept_token_count": 572.3671875, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9813625991344452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.280251738848165e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1664421081543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0453796386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16090, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.30034353733062746, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.011637707194313406, + "grad_norm": 0.10301019996404648, + "kept_token_count": 572.496875, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9812665462493897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.317776326090097e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.89681015014648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5851165771484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16100, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29940530359745027, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.01385484477505088, + "grad_norm": 0.09616143256425858, + "kept_token_count": 573.1296875, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9810958981513977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.384435521089472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17102279663087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7835357666016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16110, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.29998342990875243, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.01356473620980978, + "grad_norm": 0.1033744364976883, + "kept_token_count": 572.8375, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9814975798130036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.227526439237408e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.36661529541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3935607910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16120, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.30079675912857057, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.01412890972569585, + "grad_norm": 0.11971402913331985, + "kept_token_count": 572.0671875, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9814610660076142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.241794955916702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.98041534423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0015594482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16130, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2983655512332916, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.014928573742508888, + "grad_norm": 0.09322359412908554, + "kept_token_count": 574.071875, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9814051687717438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.26362326531671e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99746170043946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8530975341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16140, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.300255286693573, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.014073645137250424, + "grad_norm": 0.1081656739115715, + "kept_token_count": 572.2234375, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9814071238040925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.262864237418398e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.15730895996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5311553955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16150, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.3010007292032242, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.01299421014264226, + "grad_norm": 0.09427796304225922, + "kept_token_count": 571.74375, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9823505043983459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.894356265547685e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.01213302612305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9644256591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16160, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.2999379485845566, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.014037824608385563, + "grad_norm": 0.1149401143193245, + "kept_token_count": 572.9703125, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.981559818983078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.203218920039944e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.22173919677735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9169464111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.3003209292888641, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.015072250086814166, + "grad_norm": 0.13465715944766998, + "kept_token_count": 572.575, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9810917913913727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.386032593785785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.23552169799805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.35809631347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16180, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3004779189825058, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.015004715044051408, + "grad_norm": 0.15989898145198822, + "kept_token_count": 572.4046875, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9821892023086548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.957360747037456e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.37095260620117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8197479248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16190, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29927506148815153, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.016632115840911864, + "grad_norm": 0.08102796226739883, + "kept_token_count": 573.275, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9807868242263794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.505167013732716e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.73979110717774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4537719726562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16200, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.3005691885948181, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.015527860727161169, + "grad_norm": 0.11310852319002151, + "kept_token_count": 572.3234375, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.980924540758133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.451368219335563e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.44622650146485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5893524169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16210, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.299252986907959, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.0157506313174963, + "grad_norm": 0.13094662129878998, + "kept_token_count": 573.340625, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9820025563240051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.030272099655122e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.58559341430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5667358398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16220, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.29988532364368437, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.012424649205058813, + "grad_norm": 0.07702728360891342, + "kept_token_count": 572.8390625, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9819733023643493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.041701537673362e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.91074447631836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9012145996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.29966873228549956, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.011327686719596386, + "grad_norm": 0.0863189622759819, + "kept_token_count": 573.253125, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9820363998413086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.017051611910575e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.96820068359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.76259155273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16240, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.3014658480882645, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.015974353486672043, + "grad_norm": 0.11067015677690506, + "kept_token_count": 571.69375, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9818860113620758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.075790490489453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.29326934814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6706268310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16250, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.3007102757692337, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.014577564364299179, + "grad_norm": 0.10970120877027512, + "kept_token_count": 572.084375, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9809922277927399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.424925242958125e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.17306976318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.43361206054686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16260, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.3001583844423294, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.01618966441601515, + "grad_norm": 0.13168856501579285, + "kept_token_count": 572.8140625, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9817950606346131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.111317536327988e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.2678726196289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5313690185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16270, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.30080097913742065, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.014532637502998113, + "grad_norm": 0.09584730118513107, + "kept_token_count": 572.240625, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9810835659503937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.389251550193876e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2955223083496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1932861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16280, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2995254397392273, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.019038305431604386, + "grad_norm": 0.10101708024740219, + "kept_token_count": 573.140625, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9812707364559173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.316137416637502e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.23604583740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.17470092773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2993963032960892, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.015599038079380989, + "grad_norm": 0.09870662540197372, + "kept_token_count": 573.5328125, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9818349778652191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.095724358805456e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.8236457824707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9960968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16300, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.29904845058918, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.013637046981602906, + "grad_norm": 0.12235567718744278, + "kept_token_count": 573.4046875, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9821363985538483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.97798481269274e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.87595977783204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.41773376464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16310, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.29925058484077455, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.01473320173099637, + "grad_norm": 0.11555000394582748, + "kept_token_count": 573.5171875, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9810467302799225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.403639392578043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.00042343139648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6539031982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16320, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.3010427296161652, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.013360157795250415, + "grad_norm": 0.07973349094390869, + "kept_token_count": 571.984375, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9822527885437011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.93252295604907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09244079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.28443908691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16330, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2990671843290329, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.015857537183910607, + "grad_norm": 0.09591865539550781, + "kept_token_count": 573.453125, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9811540246009827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.361726893577725e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.22030487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3969360351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16340, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2996849626302719, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.013590278662741184, + "grad_norm": 0.11398613452911377, + "kept_token_count": 573.0171875, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9817248106002807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.138766450225376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.69711074829101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0330108642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16350, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2998486489057541, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.01545471465215087, + "grad_norm": 0.16427983343601227, + "kept_token_count": 572.8359375, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9806189239025116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.570758098154328e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.60007095336914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6941772460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16360, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.30047749280929564, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.01293760510161519, + "grad_norm": 0.08398132026195526, + "kept_token_count": 572.39375, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9820071637630463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.02846678905189e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.46728134155273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4083526611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16370, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.3004795640707016, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.013817932549864053, + "grad_norm": 0.10739380866289139, + "kept_token_count": 572.353125, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9811660528182984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.357030553976074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.89050674438477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.50534973144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16380, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3006328970193863, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.016140345856547354, + "grad_norm": 0.10584697127342224, + "kept_token_count": 572.234375, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9804761707782745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.626511214766651e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.94896850585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3326385498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16390, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2989310443401337, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.013284003920853138, + "grad_norm": 0.10443447530269623, + "kept_token_count": 573.58125, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9814309895038604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.25354788301047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.33655090332032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0293334960937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16400, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2998929798603058, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.013619210664182901, + "grad_norm": 0.08465347439050674, + "kept_token_count": 572.9046875, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9821707606315613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.964569183764979e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90013732910157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1078399658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16410, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2988739997148514, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.013765457086265087, + "grad_norm": 0.08019834756851196, + "kept_token_count": 573.7203125, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9810463726520539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.403779454762116e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.129638671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.2996893018484116, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.014318322483450174, + "grad_norm": 0.09529758989810944, + "kept_token_count": 573.2, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9815716624259949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.198589373729192e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.66012420654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0269378662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16430, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.30010569393634795, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.012735695019364357, + "grad_norm": 0.07878389209508896, + "kept_token_count": 572.8046875, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9828664481639862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.692813512927387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.96362686157227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.53985595703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16440, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.30057640075683595, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.014945064578205347, + "grad_norm": 0.0894356369972229, + "kept_token_count": 572.2671875, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9815753400325775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.197154773166403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.29034576416015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2195648193359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16450, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29875045716762544, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.01179264048114419, + "grad_norm": 0.10467927902936935, + "kept_token_count": 574.0046875, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9822925090789795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.917008940945379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.86424026489257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4217987060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16460, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.3004683434963226, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.013208127208054065, + "grad_norm": 0.07813028991222382, + "kept_token_count": 572.5875, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9818487048149109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.0903729647398e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.60025634765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.95607604980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16470, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.2997672200202942, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.016847529634833335, + "grad_norm": 0.1181831806898117, + "kept_token_count": 572.95625, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9814249157905579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.255906020873227e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5986343383789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3584838867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16480, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29966357052326204, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.014854324329644442, + "grad_norm": 0.12579374015331268, + "kept_token_count": 573.1640625, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9820463001728058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.013181930233258e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.39663619995117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8887969970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16490, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3001121163368225, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.011142306262627244, + "grad_norm": 0.11019536107778549, + "kept_token_count": 572.809375, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9826570987701416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.774589746783022e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.53421173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.04990539550784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16500, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.30001147389411925, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.014955665357410908, + "grad_norm": 0.14384013414382935, + "kept_token_count": 572.9078125, + "learning_rate": 1.374648002850397e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9806406378746033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.562275786767713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.95411224365235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.40410766601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16510, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.29993173480033875, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.013245510775595904, + "grad_norm": 0.11319372057914734, + "kept_token_count": 572.765625, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9818067491054535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.106753255357035e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76114959716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5806610107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16520, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30028955936431884, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.014070136938244104, + "grad_norm": 0.1433914452791214, + "kept_token_count": 572.5984375, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9819251477718354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.060510324663483e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.88805541992187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5982360839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16530, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3002885550260544, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.012408474646508693, + "grad_norm": 0.08043648302555084, + "kept_token_count": 572.571875, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9817488789558411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.129363511921838e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.11928100585938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7091400146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.29907454252243043, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.013971613673493265, + "grad_norm": 0.0843958780169487, + "kept_token_count": 573.684375, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9818777680397034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.079012066242285e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.64868392944337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1019653320312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16550, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3003185272216797, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.013580245431512594, + "grad_norm": 0.08529939502477646, + "kept_token_count": 572.565625, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9815371751785278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.212057244032621e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.68252868652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.712255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3002501159906387, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.014145010150969028, + "grad_norm": 0.08069228380918503, + "kept_token_count": 572.7171875, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.980810695886612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.495843310607597e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.33738555908204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.11431274414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16570, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3001049131155014, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.015335915517061949, + "grad_norm": 0.11904364079236984, + "kept_token_count": 573.2359375, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.98165922164917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.164389899116941e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.0612190246582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7111114501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16580, + "valid_token_count": 819.025 + }, + { + "actual_token_mask_ratio": 0.30106010735034944, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.014646837301552296, + "grad_norm": 0.10100703686475754, + "kept_token_count": 571.75, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9819490313529968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.051180764392485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.00006713867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7555694580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16590, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29929491579532624, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.016630905168130993, + "grad_norm": 0.10064077377319336, + "kept_token_count": 573.6, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9809710800647735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.43319200410042e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.13920440673829, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.00123291015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16600, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.30114589631557465, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.01404143702238798, + "grad_norm": 0.12495366483926773, + "kept_token_count": 571.890625, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9821453928947449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.974477983021644e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.48899688720704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.083154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16610, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3002963066101074, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.01446395255625248, + "grad_norm": 0.0898832380771637, + "kept_token_count": 572.4390625, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9800157070159912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.806384674040601e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.00048446655273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4797821044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16620, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3000465393066406, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.013420897722244262, + "grad_norm": 0.09089919924736023, + "kept_token_count": 572.875, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9821149468421936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.986372573010158e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.71250228881836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6413879394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16630, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.2996814578771591, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.01654992261901498, + "grad_norm": 0.1318378746509552, + "kept_token_count": 573.2328125, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9816311597824097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.175346254371107e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6095458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.76487121582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16640, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29963704347610476, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.013754553720355034, + "grad_norm": 0.154610276222229, + "kept_token_count": 573.0296875, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9815718412399292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.198521634563804e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.98513412475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.614501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16650, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3001033902168274, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.014610812254250049, + "grad_norm": 0.12319179624319077, + "kept_token_count": 572.7953125, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9817374765872955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.133815452107229e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.55598602294921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2335601806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16660, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.2997480034828186, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.016258737724274396, + "grad_norm": 0.0871797576546669, + "kept_token_count": 573.0375, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9812355101108551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.3298963252455e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.31865005493164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.48585205078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30068071484565734, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.014524803776293994, + "grad_norm": 0.12281142920255661, + "kept_token_count": 572.4265625, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9818386256694793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.094307366060094e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.86296691894532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9829467773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16680, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.29935469925403596, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.017452299874275923, + "grad_norm": 0.10637357831001282, + "kept_token_count": 573.415625, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9816537737846375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.166509240050801e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.31053771972657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.227685546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16690, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.30064129531383516, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.013787415809929371, + "grad_norm": 0.09218236804008484, + "kept_token_count": 572.4296875, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9818770408630371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.079297720338217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.58691329956055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6670288085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16700, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.29971536099910734, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.012130863778293133, + "grad_norm": 0.09709430485963821, + "kept_token_count": 573.1109375, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9822647631168365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.927837457624265e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.44024124145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.51036071777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16710, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2998771071434021, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.013133266242220998, + "grad_norm": 0.10344918072223663, + "kept_token_count": 572.9171875, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9817778050899506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.118061839719302e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.85424041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.58965759277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16720, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2998208463191986, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.013833127124235034, + "grad_norm": 0.13665886223316193, + "kept_token_count": 572.9234375, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9809932053089142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.424547147820704e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.65740051269532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8681243896484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16730, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30028578639030457, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.012943181488662959, + "grad_norm": 0.09506445378065109, + "kept_token_count": 572.7890625, + "learning_rate": 1.336795567185138e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9821874678134919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.958045123610646e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.42067108154296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.53948364257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16740, + "valid_token_count": 818.609375 + }, + { + "actual_token_mask_ratio": 0.30032619535923005, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.01563694290816784, + "grad_norm": 0.10708189755678177, + "kept_token_count": 572.603125, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9821485757827759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.973229537834413e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.78108215332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5403564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16750, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.30006697177886965, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.01336162919178605, + "grad_norm": 0.12400811165571213, + "kept_token_count": 573.021875, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9822232007980347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.944084489077796e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.53163528442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1923767089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16760, + "valid_token_count": 818.6890625 + }, + { + "actual_token_mask_ratio": 0.2995196580886841, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.014581056404858827, + "grad_norm": 0.11112358421087265, + "kept_token_count": 573.5234375, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9822680175304412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.926581663719844e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.71599578857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3533020019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16770, + "valid_token_count": 818.7546875 + }, + { + "actual_token_mask_ratio": 0.30099910497665405, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.013966088742017746, + "grad_norm": 0.11092859506607056, + "kept_token_count": 572.0984375, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9827086389064789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.754455753252842e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.1675796508789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5522186279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16780, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3000823974609375, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.013509205495938658, + "grad_norm": 0.07142084836959839, + "kept_token_count": 572.6453125, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9820577144622803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.008724278421142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.75146026611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6655639648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16790, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29871146380901337, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.015009708423167466, + "grad_norm": 0.11181570589542389, + "kept_token_count": 573.9484375, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9820601165294647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.007787571637892e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.52566146850586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.60792541503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16800, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.29951465129852295, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.014412777032703161, + "grad_norm": 0.13714782893657684, + "kept_token_count": 573.165625, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9828709363937378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.691059134027455e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.33520431518555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.42376098632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2990423977375031, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.014672117680311203, + "grad_norm": 0.10472152382135391, + "kept_token_count": 573.5078125, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9819503426551819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.050667954899836e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.96178359985352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9978363037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16820, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3006661385297775, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.016018224973231555, + "grad_norm": 0.13497136533260345, + "kept_token_count": 572.196875, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9813381910324097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.289783607120626e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.17413330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1831085205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16830, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2997356116771698, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.013076040055602789, + "grad_norm": 0.1420772522687912, + "kept_token_count": 573.315625, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.982359117269516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.89098673319677e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.34423370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5441558837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16840, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.30041159093379977, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.013691272120922804, + "grad_norm": 0.09509099274873734, + "kept_token_count": 572.5390625, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.981445038318634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.24805322533939e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09575653076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4671356201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.3003874599933624, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.013666891772300006, + "grad_norm": 0.12233678251504898, + "kept_token_count": 572.440625, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9803319811820984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.682843061047606e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.36457977294921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7559753417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16860, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3001347124576569, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.014174862625077366, + "grad_norm": 0.11368417739868164, + "kept_token_count": 572.7234375, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.982084333896637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.998324242886156e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.50023498535157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8372497558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16870, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3000716805458069, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.01295758755877614, + "grad_norm": 0.09240677207708359, + "kept_token_count": 572.89375, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9814564287662506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.243608997669071e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.84828720092773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.62465515136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16880, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.3007333904504776, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.012763021700084209, + "grad_norm": 0.08481965959072113, + "kept_token_count": 572.340625, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9830670952796936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614439880650024e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.00370864868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2346221923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16890, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.30027627348899844, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.015010574785992502, + "grad_norm": 0.10154145956039429, + "kept_token_count": 572.3828125, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9819242119789123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.060870157147293e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.07163009643554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.27753295898435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16900, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.29973338544368744, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.01234660567715764, + "grad_norm": 0.09228692948818207, + "kept_token_count": 573.3390625, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.983010220527649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.63665181491524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.9921974182129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0847869873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16910, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.3003699779510498, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.013483001198619605, + "grad_norm": 0.13286149501800537, + "kept_token_count": 572.521875, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9826668083667756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.770800609956495e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.89173812866211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7863006591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16920, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30059186518192293, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.01297349906526506, + "grad_norm": 0.0978001281619072, + "kept_token_count": 572.396875, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9816919207572937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.151610698201694e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.83163681030274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5413879394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16930, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2996542930603027, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.012299162056297063, + "grad_norm": 0.11211895197629929, + "kept_token_count": 573.1359375, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9814438045024871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.248530091601424e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.76521224975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6127899169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16940, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3003659337759018, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.013351926300674676, + "grad_norm": 0.11342952400445938, + "kept_token_count": 572.5296875, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9819678127765655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.043838522804435e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.62708892822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.00703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16950, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2991571992635727, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.011694797873497009, + "grad_norm": 0.09735343605279922, + "kept_token_count": 573.4734375, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9821142375469207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.986649095779285e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.58393020629883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7460021972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16960, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.29966062009334565, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.013244972843676806, + "grad_norm": 0.06755734980106354, + "kept_token_count": 573.215625, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9813286423683166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.29351675545331e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.66161270141602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41221313476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16970, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.3002059727907181, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.014272829424589873, + "grad_norm": 0.1167604997754097, + "kept_token_count": 572.4984375, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9815634965896607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.201776970759966e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.86946487426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.29070434570315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16980, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.2995463818311691, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.015322932554408909, + "grad_norm": 0.11438815295696259, + "kept_token_count": 573.20625, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9817344784736634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.134984407457523e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27444610595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.14971313476565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16990, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29998122751712797, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.014753931108862162, + "grad_norm": 0.10377371311187744, + "kept_token_count": 572.7078125, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9820091843605041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.027684405329638e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.2336212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.93198852539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17000, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.30011071264743805, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.015361684467643499, + "grad_norm": 0.09653881937265396, + "kept_token_count": 572.5875, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9811877906322479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.34854067559354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.4804443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17010, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30071328282356263, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.01368828946724534, + "grad_norm": 0.10723201930522919, + "kept_token_count": 572.253125, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9812803626060486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.312378693313804e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.909814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7065124511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17020, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2996362805366516, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.01283148219808936, + "grad_norm": 0.09183433651924133, + "kept_token_count": 573.00625, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9828018426895142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.718048971379175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.62984848022461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7759124755859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17030, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.299258154630661, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.014217682648450136, + "grad_norm": 0.09544862061738968, + "kept_token_count": 573.3125, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9814827084541321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.233332144096494e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36799850463868, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0793029785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17040, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2997062116861343, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.013215167447924614, + "grad_norm": 0.15359380841255188, + "kept_token_count": 572.978125, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9822109937667847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.948852460482157e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.82592849731445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6666290283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17050, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.3002650588750839, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014574917219579219, + "grad_norm": 0.09236113727092743, + "kept_token_count": 572.4421875, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9818889498710632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.074642708175816e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2623420715332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0310852050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17060, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3009155958890915, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.012505055125802755, + "grad_norm": 0.10304901003837585, + "kept_token_count": 572.0, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9807206153869629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.53103151510004e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.33662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5382141113281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17070, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29923693239688876, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.014595243986696005, + "grad_norm": 0.1055305004119873, + "kept_token_count": 573.4734375, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9824583470821381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.852229780633934e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.6223129272461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.03590087890626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.29997889399528505, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.015217268746346236, + "grad_norm": 0.09011378884315491, + "kept_token_count": 572.709375, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9824922919273377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.83896396367345e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.72020416259765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5858123779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17090, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.29962462186813354, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.013027332816272974, + "grad_norm": 0.10544206947088242, + "kept_token_count": 572.9109375, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9821640074253082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.96720031555742e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.7759895324707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.38099365234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.29978644847869873, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.016089441161602734, + "grad_norm": 0.10547371953725815, + "kept_token_count": 572.9015625, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9826036214828491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.795481676817872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.92413635253907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2215209960938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17110, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3001799464225769, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.01217316840775311, + "grad_norm": 0.10134252905845642, + "kept_token_count": 572.7515625, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9820189893245697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.023854122962802e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96300659179687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9762329101562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17120, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.3003514736890793, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.014638091996312141, + "grad_norm": 0.08194174617528915, + "kept_token_count": 572.5890625, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9817113220691681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.144040355342441e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.22761840820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.12626953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17130, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3003056675195694, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015525225549936295, + "grad_norm": 0.10899411141872406, + "kept_token_count": 572.253125, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9829102158546448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.675715194432996e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38651733398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7544738769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17140, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.29908703863620756, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.013166141044348479, + "grad_norm": 0.09741761535406113, + "kept_token_count": 573.2859375, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9823184609413147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.90687622409314e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.54659957885742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.74602966308595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17150, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.2993285357952118, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.013420998863875867, + "grad_norm": 0.07292623072862625, + "kept_token_count": 573.425, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9828970432281494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.680858386971522e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.53365707397461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6822082519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17160, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2999898999929428, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.013485896401107311, + "grad_norm": 0.12686510384082794, + "kept_token_count": 572.690625, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9821849644184113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.959014499443583e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.19600372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.65733947753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17170, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2998857319355011, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.015354679897427559, + "grad_norm": 0.14051155745983124, + "kept_token_count": 572.9453125, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9829290747642517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.668350397376344e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.80646667480468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1564239501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17180, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.30027045905590055, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013340184558182954, + "grad_norm": 0.10802685469388962, + "kept_token_count": 572.51875, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9829277038574219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66888699925039e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.84273223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.60732421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17190, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.298857182264328, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.013079814705997705, + "grad_norm": 0.08863633126020432, + "kept_token_count": 573.546875, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9816737353801728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.158712687669322e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.93831253051758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.72760620117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17200, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3006143242120743, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.013076764717698097, + "grad_norm": 0.09880008548498154, + "kept_token_count": 572.4078125, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9821690142154693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.965244610910304e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.83857116699218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.17116394042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17210, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29912649989128115, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.01595653695985675, + "grad_norm": 0.11702269315719604, + "kept_token_count": 573.3578125, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9816154599189758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.181476321420633e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9160057067871, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9342346191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17220, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.30042579770088196, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.014869039226323367, + "grad_norm": 0.10371871292591095, + "kept_token_count": 572.378125, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9817263841629028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.138149849197361e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.55117874145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5174591064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17230, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.3005324572324753, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.013012113189324737, + "grad_norm": 0.09529158473014832, + "kept_token_count": 572.671875, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9818255603313446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.099402573658153e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.73963623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3676239013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17240, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.3001652419567108, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.014249431993812322, + "grad_norm": 0.09823855012655258, + "kept_token_count": 572.6828125, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9828631043434143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.69411376293283e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61626739501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3644683837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17250, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2999472200870514, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.016178939118981363, + "grad_norm": 0.11493819206953049, + "kept_token_count": 572.6609375, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9816812336444855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.15578211384127e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.81008453369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62889099121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17260, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.30026641488075256, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.012954075075685977, + "grad_norm": 0.1194501519203186, + "kept_token_count": 572.753125, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9809816896915435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.429048309859354e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.3825698852539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.70128173828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17270, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.29952040016651155, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.013519820664077998, + "grad_norm": 0.07892126590013504, + "kept_token_count": 573.078125, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9836420774459839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.389829177351203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39099807739258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.31165771484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30009266436100007, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.01322438707575202, + "grad_norm": 0.08070026338100433, + "kept_token_count": 572.3296875, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9819566428661346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.048215229588095e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.6768814086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7931182861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17290, + "valid_token_count": 817.721875 + }, + { + "actual_token_mask_ratio": 0.3000422686338425, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.01353401830419898, + "grad_norm": 0.08673962950706482, + "kept_token_count": 572.884375, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9824913024902344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.83935133565683e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.77665710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4457580566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17300, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.2998572617769241, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.015498732496052981, + "grad_norm": 0.10149527341127396, + "kept_token_count": 573.0109375, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9812945663928986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.306828701985069e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.83608779907226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.06336364746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17310, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2997670382261276, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.014461284596472978, + "grad_norm": 0.1003599613904953, + "kept_token_count": 572.7484375, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9821506440639496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.972419760131743e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.0479736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20493469238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17320, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.3000756114721298, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.015666432306170465, + "grad_norm": 0.10345709323883057, + "kept_token_count": 572.8046875, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9820096433162689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.027499596006237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.08151779174804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41832275390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17330, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30053857564926145, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.012133321911096572, + "grad_norm": 0.08918905258178711, + "kept_token_count": 572.2921875, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9814412236213684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.249539485201239e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.46717529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.043359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17340, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2990579456090927, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.0136681760661304, + "grad_norm": 0.10104928910732269, + "kept_token_count": 573.521875, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9821040451526641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.99062722560484e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.06429824829101, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7288391113281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2998409777879715, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.012157917022705078, + "grad_norm": 0.10389965772628784, + "kept_token_count": 573.0234375, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.982495242357254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.837820692453533e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.53730697631836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3797210693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17360, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.3006096541881561, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.014574314001947641, + "grad_norm": 0.09219790995121002, + "kept_token_count": 572.4078125, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9815100967884064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.222646891023033e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.0998664855957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0817932128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17370, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3004720896482468, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.012932821176946163, + "grad_norm": 0.10267725586891174, + "kept_token_count": 572.5671875, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9825019478797913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.835195708845276e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.62758407592773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20575256347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17380, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.299289271235466, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.01370181068778038, + "grad_norm": 0.08728928118944168, + "kept_token_count": 573.15, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9817401051521302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.132793580240104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.1854118347168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4326232910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17390, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.3000208228826523, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.015915468335151672, + "grad_norm": 0.12242293357849121, + "kept_token_count": 572.515625, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9819570660591126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.048042498354334e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.58596572875976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4423858642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17400, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.3000641971826553, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.013426874950528144, + "grad_norm": 0.09353437274694443, + "kept_token_count": 572.71875, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9816967189311981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.149740267777815e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.17523956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2886657714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.2990568518638611, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.013225650787353516, + "grad_norm": 0.11292383819818497, + "kept_token_count": 573.3609375, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9805392444133758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.601872130180709e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.33700714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0425750732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17420, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.30046825110912323, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.013166990131139755, + "grad_norm": 0.13120904564857483, + "kept_token_count": 572.5828125, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9816264033317565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.1772016963223e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.08498001098633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4658477783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17430, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.3005553334951401, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.011101703578606247, + "grad_norm": 0.09330020099878311, + "kept_token_count": 572.428125, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9817922174930572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.112439634511247e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.40616912841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9846618652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17440, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30085296034812925, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.011587190395221113, + "grad_norm": 0.12346766889095306, + "kept_token_count": 572.0109375, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9836752533912658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.3768735708436e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.79776992797852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7450378417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17450, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3001271367073059, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.015360402315855027, + "grad_norm": 0.1387016773223877, + "kept_token_count": 572.55625, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9817231059074402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.139435838325881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.22152709960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1747192382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17460, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30009197294712064, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.014187122136354447, + "grad_norm": 0.10319823771715164, + "kept_token_count": 572.8421875, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9811653554439544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.357303111348301e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.97952270507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7588775634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17470, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.3007529228925705, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.0141033336520195, + "grad_norm": 0.10320647805929184, + "kept_token_count": 572.0515625, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9818277657032013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.098549613147043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.4041648864746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8515869140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17480, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.29967966079711916, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.014870123099535703, + "grad_norm": 0.11998002976179123, + "kept_token_count": 572.9921875, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9812563478946685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.321755401790142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.36788406372071, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4541839599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17490, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.299814909696579, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.012136154063045978, + "grad_norm": 0.10814539343118668, + "kept_token_count": 572.859375, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9828117072582245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.714193950756453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.57410888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.203466796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17500, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3001924782991409, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.012166907452046872, + "grad_norm": 0.11626532673835754, + "kept_token_count": 572.309375, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9826413094997406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.780754993087612e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3529769897461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2553985595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17510, + "valid_token_count": 817.8109375 + }, + { + "actual_token_mask_ratio": 0.30003662109375, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.015850839857012033, + "grad_norm": 0.10332497209310532, + "kept_token_count": 573.015625, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9808616936206818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.475922757294029e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.46293182373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9835968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17520, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.29939726889133456, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.01597512951120734, + "grad_norm": 0.0936456248164177, + "kept_token_count": 573.1921875, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9825105428695678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831835053162649e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.04959716796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1039794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17530, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2998917162418365, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.015450653154402971, + "grad_norm": 0.08363848179578781, + "kept_token_count": 572.8796875, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9819534897804261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.04943791788537e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.62552490234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.14463195800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.29920909106731414, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.013745257304981351, + "grad_norm": 0.12867528200149536, + "kept_token_count": 573.1171875, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9821931838989257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.955810458748601e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.00969467163085, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8300720214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17550, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.3004940390586853, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.013846826739609241, + "grad_norm": 0.07987924665212631, + "kept_token_count": 572.234375, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9812015950679779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.343149336520582e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.40650100708008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2993408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17560, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.2998523980379105, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.014533747266978026, + "grad_norm": 0.07400734722614288, + "kept_token_count": 573.1328125, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9822893738746643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.918232938915025e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.9862678527832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4162872314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17570, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.299375981092453, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.01468607564456761, + "grad_norm": 0.10531532764434814, + "kept_token_count": 573.046875, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.981994903087616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.033259644231293e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.31885604858398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8286590576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17580, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.2998503834009171, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.014959846623241901, + "grad_norm": 0.11628105491399765, + "kept_token_count": 572.8625, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9826136231422424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.791577361582312e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83528900146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.97328186035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17590, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2999317109584808, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.01364753358066082, + "grad_norm": 0.09010627120733261, + "kept_token_count": 572.75, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9820309996604919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.019156000751537e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.01849594116212, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2355560302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17600, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.29998327791690826, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.012663425132632255, + "grad_norm": 0.08991999179124832, + "kept_token_count": 572.903125, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9828854739665985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.685377811663785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.75537414550782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.32594604492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17610, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.30062359273433686, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.011770050600171089, + "grad_norm": 0.10937519371509552, + "kept_token_count": 572.375, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9815425932407379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.209944815258495e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.29162826538087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1759979248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3001538455486298, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.01647817958146334, + "grad_norm": 0.09969235211610794, + "kept_token_count": 572.8515625, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.981338632106781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.289619024959393e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.42813034057617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6301605224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17630, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.3002193719148636, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.015329235326498747, + "grad_norm": 0.11717721819877625, + "kept_token_count": 572.7046875, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9820174813270569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.024443184491246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.19431991577149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1619354248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17640, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3003504604101181, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.014453071262687444, + "grad_norm": 0.08453750610351562, + "kept_token_count": 572.4765625, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9820835173130036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.99864307534881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.16810455322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.56764221191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17650, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.30117549300193786, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.014440084528177976, + "grad_norm": 0.07584268599748611, + "kept_token_count": 572.090625, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9812175095081329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.336930284509435e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90666351318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.44264831542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17660, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.3005591154098511, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.011320394556969404, + "grad_norm": 0.08141592890024185, + "kept_token_count": 572.3203125, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9809758961200714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.431308840750716e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6192413330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7057678222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17670, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.299852991104126, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.012748394627124071, + "grad_norm": 0.10942022502422333, + "kept_token_count": 572.7828125, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.981569904088974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.199277206382249e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.47206039428711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9220855712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17680, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.3001000374555588, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.014977116882801057, + "grad_norm": 0.08900690823793411, + "kept_token_count": 572.871875, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.015, + "mgd_cos_sim": 0.981869226694107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.082346128299833e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.57601318359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.66289672851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17690, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.29977765679359436, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.013273568637669087, + "grad_norm": 0.10455840826034546, + "kept_token_count": 572.909375, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9812129318714142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.338716968661175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.61519241333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2859252929687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17700, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.30005423724651337, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.012451328802853823, + "grad_norm": 0.08592268824577332, + "kept_token_count": 572.6046875, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9812451422214508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.326131162699312e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1680519104004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.80808410644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17710, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2998173087835312, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.013614513259381057, + "grad_norm": 0.0863252654671669, + "kept_token_count": 573.0984375, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9821966350078583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.954452328500338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.0715560913086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9781860351562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17720, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.30063719153404234, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.013363792840391398, + "grad_norm": 0.08451887965202332, + "kept_token_count": 572.3828125, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9821902751922608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.95694896421628e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.6450180053711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.20301208496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17730, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.3001562535762787, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.012505195569247007, + "grad_norm": 0.11121083050966263, + "kept_token_count": 572.6578125, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9819966971874237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.032558569335379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.06802368164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8316192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17740, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2991427928209305, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.013865741714835168, + "grad_norm": 0.11588189750909805, + "kept_token_count": 573.6890625, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9822292625904083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.941711835679598e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44780120849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6278350830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17750, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.2996429979801178, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.011892738007009029, + "grad_norm": 0.09151214361190796, + "kept_token_count": 573.1609375, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9822349071502685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.939510385564062e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.48362579345704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.95127563476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17760, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2989360749721527, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.013014368340373039, + "grad_norm": 0.11634281277656555, + "kept_token_count": 573.6375, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9819898247718811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.035242379060946e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.49526672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1948516845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17770, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.30104524493217466, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.013679127767682075, + "grad_norm": 0.13592146337032318, + "kept_token_count": 572.0015625, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9825326442718506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.823202857049182e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.5235694885254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4676910400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17780, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2993942528963089, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.013289602473378181, + "grad_norm": 0.08349866420030594, + "kept_token_count": 573.1953125, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9828608155250549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.69501670927275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.72944412231445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8760650634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17790, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.30075951516628263, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.013398091215640307, + "grad_norm": 0.08471816778182983, + "kept_token_count": 572.2859375, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9817385792732238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.133387043722905e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.10039901733398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6889343261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17800, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.30170063078403475, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.014159105066210032, + "grad_norm": 0.10425430536270142, + "kept_token_count": 571.3828125, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.982580816745758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.804385557188653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.90750579833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62408142089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3002664834260941, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.015446681529283524, + "grad_norm": 0.1073099747300148, + "kept_token_count": 572.3671875, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9820403933525086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.015487535682041e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.00916442871093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8540466308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17820, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.30052714943885805, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.01335748117417097, + "grad_norm": 0.13472633063793182, + "kept_token_count": 572.3421875, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9819455206394195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05254809872713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.53604507446289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4389221191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17830, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3002932667732239, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.014674886874854565, + "grad_norm": 0.09378158301115036, + "kept_token_count": 572.5609375, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9801722407341004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.745236143819056e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52042770385742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.65137634277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17840, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.3006702154874802, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.014451093971729279, + "grad_norm": 0.09905552864074707, + "kept_token_count": 572.0, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9823748707771301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.884839822305367e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.63003616333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.11077270507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17850, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.3009153187274933, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.013532678131014109, + "grad_norm": 0.12143346667289734, + "kept_token_count": 572.190625, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9824650704860687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.849599303677678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.97758636474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3990051269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17860, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.3006845355033875, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.01456375103443861, + "grad_norm": 0.113420270383358, + "kept_token_count": 572.384375, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.982457411289215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.852588267065584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.67132034301758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1520935058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17870, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29924165904521943, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013574178144335747, + "grad_norm": 0.09260281920433044, + "kept_token_count": 573.3171875, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9816541492938995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.166369250626304e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.40419006347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41916809082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17880, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.3007447421550751, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013438827777281403, + "grad_norm": 0.1025143563747406, + "kept_token_count": 572.0953125, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9824346661567688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.861476867925376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.64315567016601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.95569763183596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17890, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.3010033696889877, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012900248728692531, + "grad_norm": 0.08640918880701065, + "kept_token_count": 572.1203125, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.982498437166214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.836568318249192e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.08474349975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5905395507813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17900, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.3007275462150574, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.01250361194834113, + "grad_norm": 0.07248224318027496, + "kept_token_count": 572.2984375, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9828135013580322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.713498405588325e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.64683456420899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.75189208984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17910, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2992890626192093, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.015356148779392242, + "grad_norm": 0.12483672797679901, + "kept_token_count": 573.0421875, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9824065864086151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.87244457367342e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.47431259155273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9612121582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17920, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.29934437572956085, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.01240200838074088, + "grad_norm": 0.08431393653154373, + "kept_token_count": 573.259375, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9829608201980591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.655946490354835e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.10491027832032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7956909179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17930, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2995161771774292, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.01365376808680594, + "grad_norm": 0.08833175897598267, + "kept_token_count": 573.1046875, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9823533296585083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.893245008541272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.6128662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7235565185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17940, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30033777058124544, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.012430966831743717, + "grad_norm": 0.09940285235643387, + "kept_token_count": 572.4234375, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9819501876831055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.050724998407532e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.3804313659668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.89471740722655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17950, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29951506555080415, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.013118845503777265, + "grad_norm": 0.11181998997926712, + "kept_token_count": 573.2328125, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9817263126373291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.138178661989514e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.53470382690429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.05808715820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17960, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.30004043877124786, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.015336661227047443, + "grad_norm": 0.08260498195886612, + "kept_token_count": 572.684375, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9819797456264496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.039177435217426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.24573593139648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5214782714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2994563728570938, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.010986751783639193, + "grad_norm": 0.09182171523571014, + "kept_token_count": 573.2921875, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9814223706722259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.256904718815349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.9244155883789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6672821044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17980, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2992893934249878, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.01176252351142466, + "grad_norm": 0.1108253225684166, + "kept_token_count": 573.4796875, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9825481295585632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.81715115206316e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.2393569946289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.12177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17990, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.29952832460403445, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.01344267586246133, + "grad_norm": 0.13536523282527924, + "kept_token_count": 573.4625, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9820220470428467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.022654972388409e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.3728157043457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.97682189941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18000, + "valid_token_count": 818.6828125 + }, + { + "actual_token_mask_ratio": 0.299710288643837, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.012595856189727783, + "grad_norm": 0.13457459211349487, + "kept_token_count": 572.9765625, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9821360409259796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.978125020395964e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.56735458374024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.89402160644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.300088968873024, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.01277811974287033, + "grad_norm": 0.07654738426208496, + "kept_token_count": 572.759375, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9820821166038514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.999187462497503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.41367263793946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.11217041015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18020, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30047099888324735, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.013518241420388221, + "grad_norm": 0.11408381164073944, + "kept_token_count": 572.48125, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9819826424121857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.038044277578593e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.73985443115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8211334228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18030, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.29984056055545805, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.013801801856607199, + "grad_norm": 0.0915428176522255, + "kept_token_count": 572.803125, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9827284276485443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.746729377482552e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.08724822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9119598388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18040, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.29976747930049896, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.013337953202426433, + "grad_norm": 0.09700781106948853, + "kept_token_count": 572.9734375, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9827324092388153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.745169775967952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.50182495117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.31492309570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18050, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.299746036529541, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.01290352437645197, + "grad_norm": 0.12054217606782913, + "kept_token_count": 573.2, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9820297360420227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.019658442004584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.3095085144043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1676330566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18060, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.30002132058143616, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.013318074401468039, + "grad_norm": 0.11901490390300751, + "kept_token_count": 572.815625, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9827875435352326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.723624828737229e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5781349182129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.60802307128904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18070, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30056575536727903, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.013921273127198219, + "grad_norm": 0.09817370027303696, + "kept_token_count": 572.3109375, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9815271735191345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.215963851194828e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.50533828735351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7347839355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18080, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30020790100097655, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.014376093819737434, + "grad_norm": 0.10306137055158615, + "kept_token_count": 572.66875, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9819226503372193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.061482865537983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.2673942565918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0678955078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18090, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2990136295557022, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.01254411619156599, + "grad_norm": 0.08259400725364685, + "kept_token_count": 573.528125, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9818875432014466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.075189569150098e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.10662612915038, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4695098876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18100, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.30107594430446627, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.013333302456885576, + "grad_norm": 0.13183413445949554, + "kept_token_count": 571.825, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9819278836250305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.059433046379127e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.50613861083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2462982177734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18110, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.2998594373464584, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.013140085572376847, + "grad_norm": 0.11047059297561646, + "kept_token_count": 572.959375, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9815106093883514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.222433705464937e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27786865234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7405548095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30042884945869447, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.01246490110643208, + "grad_norm": 0.10141505300998688, + "kept_token_count": 572.5546875, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9809239864349365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.45159137295559e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.77035827636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.33763427734374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18130, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3000612109899521, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.01639980785548687, + "grad_norm": 0.0962609052658081, + "kept_token_count": 572.7125, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9823683798313141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.887367853778415e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.98440399169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.32146911621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18140, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3003297209739685, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.012185845244675874, + "grad_norm": 0.09637732058763504, + "kept_token_count": 572.409375, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9830515325069428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.620520289288834e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.1764404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4803497314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18150, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30055617690086367, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.012596763204783201, + "grad_norm": 0.08201055973768234, + "kept_token_count": 572.1546875, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.981017017364502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.415250074700452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.9405128479004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4701751708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18160, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2993344306945801, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.016206166706979275, + "grad_norm": 0.07017413526773453, + "kept_token_count": 573.321875, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9819359362125397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.056293688947334e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.80503158569336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.45314331054686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18170, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2994137227535248, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.013648002687841655, + "grad_norm": 0.08011767268180847, + "kept_token_count": 573.203125, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9817035019397735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.147087671910413e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.92082138061524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.23936462402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18180, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2998338669538498, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.013927299156785011, + "grad_norm": 0.09806975722312927, + "kept_token_count": 572.7, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9816967666149139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.149719167500734e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.5289192199707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.24903869628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18190, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.30053905248641966, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.011856747744604944, + "grad_norm": 0.07140412926673889, + "kept_token_count": 572.5984375, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9816842436790466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.154614977480377e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.434521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3664520263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18200, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.30027173161506654, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.012429299112409354, + "grad_norm": 0.11241117119789124, + "kept_token_count": 572.5625, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9820114493370056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.02679477399215e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.79234008789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.57525024414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18210, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.3003857582807541, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.015991429332643747, + "grad_norm": 0.11053597182035446, + "kept_token_count": 572.5890625, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9817483246326446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.129583100322634e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.2297752380371, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.27068481445315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18220, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2999589055776596, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.012792300200089812, + "grad_norm": 0.08791279047727585, + "kept_token_count": 572.5828125, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9819432377815247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05344406014774e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.93749923706055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.85235900878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18230, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.300016263127327, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.013102788105607032, + "grad_norm": 0.0959038957953453, + "kept_token_count": 572.5609375, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9819130003452301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.065252066240646e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.5521110534668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8749725341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18240, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2995995610952377, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.013172973319888115, + "grad_norm": 0.10096771270036697, + "kept_token_count": 572.928125, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.983006888628006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.637951664743014e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.24971466064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5730682373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29979623556137086, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.013385703414678573, + "grad_norm": 0.11097705364227295, + "kept_token_count": 572.9140625, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9821026623249054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.991168993408792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.26922988891602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.51658935546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18260, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30037164986133574, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.015167472325265408, + "grad_norm": 0.0928829237818718, + "kept_token_count": 572.478125, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9815679132938385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.200048785307444e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.33027801513671, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9604095458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18270, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.3013256102800369, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.012913880869746208, + "grad_norm": 0.1085166186094284, + "kept_token_count": 571.7375, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9829305529594421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.667766756436322e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.05792083740235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5169403076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18280, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3007041811943054, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.013767191302031278, + "grad_norm": 0.08941801637411118, + "kept_token_count": 572.0015625, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9821502685546875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.972573391976767e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.33081741333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1492065429687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18290, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.30032598078250883, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.012612442672252654, + "grad_norm": 0.12492933869361877, + "kept_token_count": 572.4484375, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9821369409561157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.977774974075146e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.04393463134765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.04673767089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18300, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.300115293264389, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.014649189263582229, + "grad_norm": 0.10531754791736603, + "kept_token_count": 572.7078125, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9827634811401367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.733030968462118e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.18775482177735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.27606201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2994213134050369, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.012823650240898132, + "grad_norm": 0.0765732005238533, + "kept_token_count": 573.278125, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.981833428144455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.096340195857919e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.99854888916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.47430725097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18320, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2987043887376785, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.016223618667572735, + "grad_norm": 0.10488015413284302, + "kept_token_count": 573.734375, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9814370095729827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.251187562360429e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.71602020263671, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.12591857910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18330, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2998009413480759, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.013397068297490478, + "grad_norm": 0.1106482520699501, + "kept_token_count": 573.10625, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.980558556318283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.594330891151913e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.91218948364258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9372894287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18340, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.29946704506874083, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.011628007236868142, + "grad_norm": 0.10051623731851578, + "kept_token_count": 573.20625, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9825688660144806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.809060359955766e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.66341857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.43633728027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18350, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3001578688621521, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.012887030048295856, + "grad_norm": 0.10124948620796204, + "kept_token_count": 572.7140625, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9817956149578094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.11110551492311e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.10781173706054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8394317626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18360, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2997567385435104, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01318858191370964, + "grad_norm": 0.11540865153074265, + "kept_token_count": 572.9328125, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9826281249523163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.785907607991249e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.48603134155273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.89130249023435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18370, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2995713442564011, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.015247238334268332, + "grad_norm": 0.1142750009894371, + "kept_token_count": 573.21875, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9823290109634399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.902756067574955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.37419662475585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.96727600097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18380, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2994255661964417, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.011992800841107964, + "grad_norm": 0.09083590656518936, + "kept_token_count": 573.4671875, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9828477144241333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.700125013594515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.93993072509765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.92225341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18390, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.3005813419818878, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.01463358774781227, + "grad_norm": 0.14272616803646088, + "kept_token_count": 572.375, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9814962923526764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.228024114738218e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.17353820800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1027099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18400, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2998049080371857, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.01480855531990528, + "grad_norm": 0.0948975533246994, + "kept_token_count": 573.31875, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9816866099834443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.153689293772913e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.57584838867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8750457763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18410, + "valid_token_count": 818.8 + }, + { + "actual_token_mask_ratio": 0.3009495884180069, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.012073294352740049, + "grad_norm": 0.14802291989326477, + "kept_token_count": 572.0265625, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9828108489513397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.71453497488983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.71246109008788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6526214599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18420, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3001044899225235, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.014657975919544698, + "grad_norm": 0.11616816371679306, + "kept_token_count": 572.775, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9819212079048156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.062044169288128e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.63005447387695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5357330322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18430, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30134742259979247, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.012642194051295519, + "grad_norm": 0.09886247664690018, + "kept_token_count": 571.8140625, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9811498284339905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.36337089620065e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34062805175782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.76943054199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18440, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.2993545889854431, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.01276044393889606, + "grad_norm": 0.13059137761592865, + "kept_token_count": 573.3765625, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9819549024105072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.048887346172705e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.09828720092773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.78615112304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18450, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3002175748348236, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.014223545463755727, + "grad_norm": 0.0676022544503212, + "kept_token_count": 572.590625, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.981391179561615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.269092166097834e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.11044158935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2649322509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18460, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30021646022796633, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.012383043672889471, + "grad_norm": 0.09860655665397644, + "kept_token_count": 572.6171875, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9821544706821441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.970928880036808e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.60238952636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8528594970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18470, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29907287657260895, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.012551161926239729, + "grad_norm": 0.08788739889860153, + "kept_token_count": 573.740625, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9824667930603027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.848928023828194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.98349075317383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.94457092285154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18480, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.2986166745424271, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.012769185099750757, + "grad_norm": 0.09368373453617096, + "kept_token_count": 573.8765625, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9813160955905914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.298420277948025e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.13118133544921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3628234863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999112635850906, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.013368351384997368, + "grad_norm": 0.07865793257951736, + "kept_token_count": 572.9234375, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9829696536064148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.652499796473421e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.77166213989258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0669219970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.299867844581604, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.014383438229560851, + "grad_norm": 0.09793955087661743, + "kept_token_count": 573.05, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.982699990272522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.757834707968869e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.99088516235352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.87322692871095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18510, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3005693107843399, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.014360387064516544, + "grad_norm": 0.11137264966964722, + "kept_token_count": 572.5796875, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.981971275806427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.042485085548834e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.17186584472657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3675140380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18520, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.30106281042099, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.011927231494337321, + "grad_norm": 0.10349592566490173, + "kept_token_count": 571.9625, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9822710394859314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.925402776687406e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.91090316772461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8802215576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18530, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2998180627822876, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.013811359461396933, + "grad_norm": 0.11502520740032196, + "kept_token_count": 573.0859375, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9826481103897095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.778107745049055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.99312896728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.44244384765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18540, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.3010676622390747, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.013057690067216753, + "grad_norm": 0.09703981131315231, + "kept_token_count": 571.925, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.982581377029419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.804167642258108e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.80575866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.05584716796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18550, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.29877784550189973, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.01332380948588252, + "grad_norm": 0.10018479079008102, + "kept_token_count": 573.6046875, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9828559875488281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.696892232866958e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.05705337524414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.76768188476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18560, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.2996410459280014, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.014301800448447467, + "grad_norm": 0.09484200179576874, + "kept_token_count": 572.765625, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9814002394676209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.265545209520497e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.20682678222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5736511230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18570, + "valid_token_count": 817.8171875 + }, + { + "actual_token_mask_ratio": 0.30026071071624755, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.011643290612846613, + "grad_norm": 0.09323768317699432, + "kept_token_count": 572.7609375, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9816206455230713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.179456370067782e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.00922927856445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7349914550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18580, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.29958140552043916, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.014248750265687704, + "grad_norm": 0.09988265484571457, + "kept_token_count": 573.2234375, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9819540619850159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.049218111205846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.31353607177735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16766967773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18590, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.3000267595052719, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.013143624551594258, + "grad_norm": 0.14061838388442993, + "kept_token_count": 572.5953125, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9821561753749848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.970264221308752e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.40271377563477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0831237792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18600, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.29923374354839327, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.014304604753851891, + "grad_norm": 0.0923743024468422, + "kept_token_count": 573.596875, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9827135860919952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.752521367161534e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.55264892578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5537414550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18610, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.30085289478302, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.01297092493623495, + "grad_norm": 0.1277395635843277, + "kept_token_count": 571.9171875, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9813697874546051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.277441000042018e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22497100830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8864501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18620, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.3005930960178375, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.012955678161233664, + "grad_norm": 0.07994439452886581, + "kept_token_count": 572.4625, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9820502877235413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.011626621533651e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.01142272949218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3964416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18630, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.300503733754158, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.012287971563637257, + "grad_norm": 0.0831536129117012, + "kept_token_count": 572.6015625, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9817362785339355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.134291154216043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9801528930664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4082733154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18640, + "valid_token_count": 818.5875 + }, + { + "actual_token_mask_ratio": 0.2998687714338303, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.01191370040178299, + "grad_norm": 0.10306467860937119, + "kept_token_count": 572.903125, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9816743493080139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.158478183555417e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.1124153137207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8226715087891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3001488626003265, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.012538661900907754, + "grad_norm": 0.10664013028144836, + "kept_token_count": 572.7640625, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9826607406139374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.773168752260972e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2195053100586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.15684204101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18660, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29956196844577787, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.010420428682118654, + "grad_norm": 0.09782067686319351, + "kept_token_count": 573.1125, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9820327043533326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.018497271928936e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.43016204833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6460418701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18670, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29955486953258514, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.013539869990199804, + "grad_norm": 0.1123233214020729, + "kept_token_count": 573.446875, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9824604034423828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.851422658655792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.68549194335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18680, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.2998418062925339, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.013869461324065923, + "grad_norm": 0.1152888610959053, + "kept_token_count": 572.69375, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9831628918647766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.577016683877446e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5985321044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2959716796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18690, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30031490325927734, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.01449607377871871, + "grad_norm": 0.08661311864852905, + "kept_token_count": 572.4640625, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9820106089115143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.027128085610457e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8802993774414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9439758300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18700, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.299998876452446, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.013067945279181004, + "grad_norm": 0.10554426908493042, + "kept_token_count": 572.88125, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9819918036460876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.034466179902665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.83386688232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5966094970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18710, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2990685164928436, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.014296845253556966, + "grad_norm": 0.11771725863218307, + "kept_token_count": 573.3609375, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9818721771240234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.08120256604161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.33897552490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3929016113281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18720, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.3006330817937851, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.011975138774141669, + "grad_norm": 0.10930171608924866, + "kept_token_count": 572.4171875, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9811906278133392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.347428036155179e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.00075912475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2855712890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18730, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.2999222457408905, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.016017176024615765, + "grad_norm": 0.10948057472705841, + "kept_token_count": 573.225, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9815446734428406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.209128598333337e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.1813980102539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.97893676757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18740, + "valid_token_count": 818.8015625 + }, + { + "actual_token_mask_ratio": 0.3001692593097687, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.013456245791167021, + "grad_norm": 0.07985781133174896, + "kept_token_count": 572.678125, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9831414759159088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.585381452168803e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.34543151855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.52564697265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18750, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2994076102972031, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.013309578085318208, + "grad_norm": 0.09319359809160233, + "kept_token_count": 573.3203125, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9821017622947693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.991520785959437e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.61478881835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47178955078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18760, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29925357401371, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.013120154477655887, + "grad_norm": 0.08788483589887619, + "kept_token_count": 573.471875, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9830384194850922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.625636233366094e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.06236190795899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1214233398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18770, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30006625652313235, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.012571771303191781, + "grad_norm": 0.07750590145587921, + "kept_token_count": 572.8046875, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9821983695030212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.953780502954033e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9033592224121, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6278533935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18780, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.30060355067253114, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.012341589247807861, + "grad_norm": 0.09911074489355087, + "kept_token_count": 572.196875, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9811540901660919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.36170077288989e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.97727279663086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2193634033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2997090607881546, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.014798876456916332, + "grad_norm": 0.10181654244661331, + "kept_token_count": 573.340625, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9811830282211303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.350399682763964e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.97642059326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4143829345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18800, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.30070637762546537, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.012207921873778105, + "grad_norm": 0.11872879415750504, + "kept_token_count": 572.26875, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9819839239120484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.037546092760749e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.4840690612793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.26764526367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18810, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2998697519302368, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.01358715733513236, + "grad_norm": 0.11594906449317932, + "kept_token_count": 573.096875, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9832066416740417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.559918947459664e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.85509643554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.28843688964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18820, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.30013165473937986, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.013346299389377236, + "grad_norm": 0.10798688977956772, + "kept_token_count": 572.553125, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9821312010288239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.980018770263996e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.78778991699218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3483428955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18830, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.2994366377592087, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.013734617177397013, + "grad_norm": 0.0821242481470108, + "kept_token_count": 573.225, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9819319605827331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05784470483195e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7146827697754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.77569885253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18840, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30004751980304717, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.014406171627342701, + "grad_norm": 0.12554465234279633, + "kept_token_count": 572.8609375, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9817832171916961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.115954213077202e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.50496368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.14451293945314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18850, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.3002193748950958, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.012513687182217837, + "grad_norm": 0.08213935047388077, + "kept_token_count": 572.4375, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9829817831516265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.647764021181502e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.0237190246582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2947296142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18860, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.30003589391708374, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.012784274853765964, + "grad_norm": 0.11299420893192291, + "kept_token_count": 572.8375, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9825107753276825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831748760305345e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.78035964965821, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4526824951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18870, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.29981821179389956, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.01198470089584589, + "grad_norm": 0.083603635430336, + "kept_token_count": 573.03125, + "learning_rate": 9.928266614677168e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9822671055793762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.926931455382146e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.98708419799804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.24993896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18880, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3001133441925049, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.012921978207305073, + "grad_norm": 0.09365836530923843, + "kept_token_count": 572.509375, + "learning_rate": 9.912708818626524e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9827789247035981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.726998544763774e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.37337036132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.55187072753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18890, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.3007448762655258, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.012514927331358195, + "grad_norm": 0.07799743115901947, + "kept_token_count": 572.2109375, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9823638021945953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.889158685225993e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82903289794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1826568603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18900, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.2989510715007782, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.011924154125154018, + "grad_norm": 0.0914163887500763, + "kept_token_count": 573.634375, + "learning_rate": 9.881611790037375e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9829989492893219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.641048385063187e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.38811340332032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3666961669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18910, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.3007592499256134, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.014377529360353946, + "grad_norm": 0.11902139335870743, + "kept_token_count": 572.2203125, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.982457447052002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.852574151707813e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.81700057983399, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4881591796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18920, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.3000160813331604, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.014007293153554201, + "grad_norm": 0.11421523243188858, + "kept_token_count": 572.7546875, + "learning_rate": 9.850539638737894e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9816437721252441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.170422613853589e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.21253051757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.60011291503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18930, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2989980161190033, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.012002446921542287, + "grad_norm": 0.10658260434865952, + "kept_token_count": 573.6625, + "learning_rate": 9.835012939271006e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9820505619049072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.011513916950207e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5681770324707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8015960693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18940, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.30053366124629977, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.013465599156916142, + "grad_norm": 0.10306808352470398, + "kept_token_count": 572.3734375, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9819378137588501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05556434695609e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.15454864501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7970764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18950, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.30037998855113984, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.01276254653930664, + "grad_norm": 0.10328279435634613, + "kept_token_count": 572.44375, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9815354287624359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.212741038529203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.92694778442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5499237060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18960, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.30021811425685885, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.015243136184290051, + "grad_norm": 0.06810537725687027, + "kept_token_count": 572.3953125, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9822207093238831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.945056520635262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.10299835205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0700653076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18970, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2993845134973526, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.01395918633788824, + "grad_norm": 0.1018528938293457, + "kept_token_count": 573.284375, + "learning_rate": 9.772969089416008e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9823940336704254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.877350315335207e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.70819931030273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6275939941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18980, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2999952256679535, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.013555200956761836, + "grad_norm": 0.13068115711212158, + "kept_token_count": 572.93125, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9821821630001069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.960106620681472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.83199996948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.18170471191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18990, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2992973655462265, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.012541353749111295, + "grad_norm": 0.12111660838127136, + "kept_token_count": 573.4015625, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9822725057601929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.924819463165477e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.7978530883789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1394958496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19000, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.30041679441928865, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.012583150528371334, + "grad_norm": 0.08941996097564697, + "kept_token_count": 572.4375, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9822723150253296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.924894041731023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.03830032348633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1686248779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19010, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29995993971824647, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.012561126053333283, + "grad_norm": 0.07860852777957916, + "kept_token_count": 572.684375, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9820047974586487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.029386761132627e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.32674560546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7532928466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19020, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.29908322393894193, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.014949298044666649, + "grad_norm": 0.12059799581766129, + "kept_token_count": 573.6015625, + "learning_rate": 9.69555732309408e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9818301141262055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.097632696968503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.17211837768555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8450439453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19030, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.29928408861160277, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.015786742325872183, + "grad_norm": 0.1253100037574768, + "kept_token_count": 573.6203125, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9824757218360901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.845442549092695e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.65257949829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.73350219726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19040, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.2995947688817978, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.01386340195313096, + "grad_norm": 0.08696338534355164, + "kept_token_count": 572.9765625, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9812501668930054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.324172765947879e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.94592590332032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5247833251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19050, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3011456996202469, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.016105829272419214, + "grad_norm": 0.1077108234167099, + "kept_token_count": 571.7953125, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9818340361118316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.096099798218348e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82923431396485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.56928100585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19060, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.3000843822956085, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.01472284346818924, + "grad_norm": 0.12077038735151291, + "kept_token_count": 572.8359375, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.981506621837616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.223992179206107e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8324821472168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7402404785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19070, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.300652003288269, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.01415335964411497, + "grad_norm": 0.10932335257530212, + "kept_token_count": 572.2546875, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.981675922870636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.157867512432858e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.24707336425782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.73387756347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19080, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.30056368708610537, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.012983838794752955, + "grad_norm": 0.12127411365509033, + "kept_token_count": 572.20625, + "learning_rate": 9.602876107719228e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9821344912052155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.978732599236537e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.35039520263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6668731689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19090, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2997813940048218, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.014551590476185084, + "grad_norm": 0.1279907524585724, + "kept_token_count": 572.9765625, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9821805894374848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.960725731914863e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57080001831055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4884399414062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19100, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.29940454065799715, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.014991295617073774, + "grad_norm": 0.09361843764781952, + "kept_token_count": 573.4875, + "learning_rate": 9.57203463328011e-06, + "loss": 0.015, + "mgd_cos_sim": 0.982435405254364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.861186993774026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39221267700195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3508636474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19110, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.29914267659187316, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.0152551282197237, + "grad_norm": 0.11218826472759247, + "kept_token_count": 573.2296875, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9814110338687897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.261325627041514e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.59089965820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.95912475585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19120, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.2994254261255264, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.014266482647508382, + "grad_norm": 0.148391991853714, + "kept_token_count": 573.5359375, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9810991525650025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.383163101621903e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.47184448242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0859130859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19130, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.3008254587650299, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.014553626254200935, + "grad_norm": 0.12155076116323471, + "kept_token_count": 571.934375, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9819911181926727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.034740083327052e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.31730270385742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4037292480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19140, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.29973708689212797, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.013959010317921638, + "grad_norm": 0.09770462661981583, + "kept_token_count": 572.9703125, + "learning_rate": 9.510430980027684e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9815871775150299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.192532357294112e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.45483322143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5421539306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19150, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30030514895915983, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.014638865645974875, + "grad_norm": 0.10308133810758591, + "kept_token_count": 572.4859375, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9812483668327332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.324875186895952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.44068222045898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3392669677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19160, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.29922678172588346, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.011138607421889902, + "grad_norm": 0.2859726846218109, + "kept_token_count": 573.7203125, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9824536263942718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.854067396488972e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91043319702149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7574401855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19170, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.3001915454864502, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.014519748277962208, + "grad_norm": 0.09561581164598465, + "kept_token_count": 572.709375, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.982653820514679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.775870351702907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.96367111206055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7800628662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19180, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29993678629398346, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.013817589823156595, + "grad_norm": 0.09125083684921265, + "kept_token_count": 572.98125, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9821393430233002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.976834702072665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.82389068603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7800994873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19190, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.30016849040985105, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.014977837214246392, + "grad_norm": 0.09870681911706924, + "kept_token_count": 572.553125, + "learning_rate": 9.433576636128284e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9822827756404877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.920808846189175e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.2450553894043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.38336486816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19200, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.3008661448955536, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.012333508487790823, + "grad_norm": 0.09376582503318787, + "kept_token_count": 571.859375, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9819407939910889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.054396555759013e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.38402862548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5801055908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19210, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.3011713266372681, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.013283495791256428, + "grad_norm": 0.08713112026453018, + "kept_token_count": 571.9328125, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.982287484407425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.918972358107566e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.74816970825195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2760375976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19220, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3008339285850525, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.014175094105303288, + "grad_norm": 0.12240270525217056, + "kept_token_count": 571.80625, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9817598223686218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.125087940949016e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.26859054565429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3579406738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19230, + "valid_token_count": 817.8375 + }, + { + "actual_token_mask_ratio": 0.30109787583351133, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.012909762328490615, + "grad_norm": 0.08114740252494812, + "kept_token_count": 571.6984375, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9833261728286743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.513228399853687e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.17501678466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4333251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19240, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.30050958395004274, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.014427577145397663, + "grad_norm": 0.11052357405424118, + "kept_token_count": 572.4546875, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9830748856067657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.61139561998425e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.79239349365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.10362243652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19250, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.29938929378986356, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.013056810852140189, + "grad_norm": 0.08887147158384323, + "kept_token_count": 573.33125, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9821630895137787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.967558583710342e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09306259155274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6111145019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19260, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2987340748310089, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.010696030873805284, + "grad_norm": 0.12045112252235413, + "kept_token_count": 574.0484375, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9826211035251617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.788648388464936e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.94174423217774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1462371826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19270, + "valid_token_count": 818.584375 + }, + { + "actual_token_mask_ratio": 0.30082741379737854, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.01647932520136237, + "grad_norm": 0.11312314122915268, + "kept_token_count": 571.8, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9817504584789276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.128745201043785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.72672119140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6545928955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19280, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.3003454953432083, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.012061258312314749, + "grad_norm": 0.1374361515045166, + "kept_token_count": 572.4109375, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9820456743240357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.013430804363452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.16640548706054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.33897094726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19290, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2990530073642731, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.014995299745351076, + "grad_norm": 0.10607153922319412, + "kept_token_count": 573.5046875, + "learning_rate": 9.280377556281306e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9814732849597931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.23701963579515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.67276992797852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0951324462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19300, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.30019075572490694, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.013553832564502955, + "grad_norm": 0.08322872966527939, + "kept_token_count": 572.778125, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9816821992397309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.155413331929594e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.53490142822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.58961791992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3003017961978912, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.015264432784169913, + "grad_norm": 0.09874662011861801, + "kept_token_count": 572.4609375, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9815934121608734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.190096330305096e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.84548645019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.02207336425784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19320, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2995529741048813, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.012182805361226201, + "grad_norm": 0.09547818452119827, + "kept_token_count": 573.2265625, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9841223657131195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.20221526332898e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.56856384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1573974609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19330, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.30034866333007815, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.012870718911290168, + "grad_norm": 0.12446673959493637, + "kept_token_count": 572.5375, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9825417816638946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.819628688390367e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.80107116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8826049804687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19340, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2985983371734619, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.011705214576795697, + "grad_norm": 0.07587471604347229, + "kept_token_count": 573.9109375, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9829841673374176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.646826768701431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.76949539184571, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.838037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.29996325373649596, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.012770812772214413, + "grad_norm": 0.10320243239402771, + "kept_token_count": 572.75, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.981803297996521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.108100180630572e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.49046478271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.74273681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19360, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.299984472990036, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.014995558932423592, + "grad_norm": 0.12087511271238327, + "kept_token_count": 572.7625, + "learning_rate": 9.173550609917186e-06, + "loss": 0.015, + "mgd_cos_sim": 0.982339757680893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.8985501275165e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.82910232543945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.20717163085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19370, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29892646968364717, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.012522982619702815, + "grad_norm": 0.07680729776620865, + "kept_token_count": 573.740625, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.982321846485138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.905543050379492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.16984176635742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.89788208007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19380, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2997501105070114, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.013867551553994417, + "grad_norm": 0.1124250665307045, + "kept_token_count": 573.14375, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.982408857345581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.871559780847747e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.1794662475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8633056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.30035252273082735, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.014267941331490874, + "grad_norm": 0.08921211212873459, + "kept_token_count": 572.6203125, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9825867772102356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.802057687309571e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.8843864440918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2497802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19400, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30015126466751096, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.013056989992037416, + "grad_norm": 0.1273857057094574, + "kept_token_count": 572.3375, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.983967524766922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.262702809181064e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.7855598449707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8532379150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19410, + "valid_token_count": 817.803125 + }, + { + "actual_token_mask_ratio": 0.29933003187179563, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.013646522723138332, + "grad_norm": 0.12231101095676422, + "kept_token_count": 573.0875, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9825481593608856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.817142202635296e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.70585784912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8267761230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19420, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.2991936832666397, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.011507755750790239, + "grad_norm": 0.150371715426445, + "kept_token_count": 573.240625, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9831724047660828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.573294194822665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.17053527832032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7218994140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19430, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.29887485802173613, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.016407044138759375, + "grad_norm": 0.11859375983476639, + "kept_token_count": 573.6421875, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9814764022827148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.235800512717105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.3473617553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9423889160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19440, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.3019264876842499, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.010466128308326005, + "grad_norm": 0.15721404552459717, + "kept_token_count": 571.25, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9828141927719116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.713227994623595e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.6381820678711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7762115478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19450, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.30021598637104036, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.012391105014830827, + "grad_norm": 0.08005981892347336, + "kept_token_count": 572.3578125, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.982569944858551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.80863293382572e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.76338119506836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6409545898438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19460, + "valid_token_count": 817.90625 + }, + { + "actual_token_mask_ratio": 0.30015318691730497, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.012458765646442771, + "grad_norm": 0.11497914046049118, + "kept_token_count": 572.5921875, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9823120534420013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.90937195031438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.87261276245117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.59081115722654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19470, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29986409842967987, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.01208298304118216, + "grad_norm": 0.09336981922388077, + "kept_token_count": 572.775, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9824015617370605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.874413011246361e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.23407516479492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.33652648925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2997114658355713, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.012388614192605018, + "grad_norm": 0.0901133194565773, + "kept_token_count": 573.3046875, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9820874512195588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.997106866037938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.96504058837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2201263427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19490, + "valid_token_count": 818.6609375 + }, + { + "actual_token_mask_ratio": 0.30014865696430204, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.01217767596244812, + "grad_norm": 0.07581479847431183, + "kept_token_count": 572.665625, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9816303789615631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.17565548256971e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.4235237121582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.59630126953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19500, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2991804271936417, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.016303099505603314, + "grad_norm": 0.09823282808065414, + "kept_token_count": 573.390625, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9829904556274414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.64437440718757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.74901275634765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.86058044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19510, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.3003265142440796, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.013017533533275127, + "grad_norm": 0.1103362962603569, + "kept_token_count": 572.33125, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9819773435592651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.040122472972144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.15063781738282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7967926025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19520, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.300573468208313, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.014940327545627952, + "grad_norm": 0.10319963097572327, + "kept_token_count": 572.309375, + "learning_rate": 8.93068758863277e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9808969140052796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.462153298547492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.89227905273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7389801025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19530, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29843064248561857, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.012208343297243119, + "grad_norm": 0.08857952058315277, + "kept_token_count": 574.1828125, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9810314238071441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.409621721308213e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.3849609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8714904785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19540, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.30133027136325835, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.013664204813539981, + "grad_norm": 0.10294122248888016, + "kept_token_count": 571.646875, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9827067852020264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.755183967470657e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.68192443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8869354248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19550, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3000877439975739, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.012801900506019592, + "grad_norm": 0.09838853776454926, + "kept_token_count": 572.8515625, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9821779906749726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.96174225595314e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.29594497680664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62400207519534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19560, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.299945193529129, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.013568704714998603, + "grad_norm": 0.09495513886213303, + "kept_token_count": 573.0171875, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9824230551719666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.866010007797741e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.66082305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6625244140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19570, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30195771157741547, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.012490305397659541, + "grad_norm": 0.08684733510017395, + "kept_token_count": 571.1296875, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9820035398006439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.02989098499529e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.27180480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.91640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19580, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2997920781373978, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011595327267423272, + "grad_norm": 0.09386661648750305, + "kept_token_count": 573.159375, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9828965485095977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.681058221147396e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.58937225341796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3307647705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19590, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.2999603092670441, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.013168784882873296, + "grad_norm": 0.11327319592237473, + "kept_token_count": 572.95, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9823975622653961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.875971230329015e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.41750869750976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2336700439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19600, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.3002185791730881, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.012551246024668217, + "grad_norm": 0.07752327620983124, + "kept_token_count": 572.6890625, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9831528186798095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.580947374459356e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.94394378662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2121826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19610, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2995326370000839, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.012648898269981146, + "grad_norm": 0.11624974012374878, + "kept_token_count": 573.0921875, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9829304337501525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.667815105174669e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.98020782470704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8699188232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19620, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.29958691000938414, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.013716216664761305, + "grad_norm": 0.11845188587903976, + "kept_token_count": 573.1875, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9824830412864685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.84258331602905e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.37384948730468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6362548828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19630, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.300419545173645, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.012571214977651835, + "grad_norm": 0.09891024231910706, + "kept_token_count": 572.4390625, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.983011531829834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.636144862568472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.4800651550293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5720458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19640, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.3003705471754074, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.013197468314319848, + "grad_norm": 0.09260274469852448, + "kept_token_count": 572.421875, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9821580529212952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.969529567868449e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.51156234741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.11668090820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19650, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3001977801322937, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.015355955343693495, + "grad_norm": 0.10024555027484894, + "kept_token_count": 572.5625, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9818434953689575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.092399173416197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61664810180665, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.02728271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19660, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.30037675201892855, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.015846274700015782, + "grad_norm": 0.10742540657520294, + "kept_token_count": 572.5921875, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.982313734292984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.908717405167409e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.01140365600585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19670, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.3007615298032761, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.012821113783866167, + "grad_norm": 0.09583509713411331, + "kept_token_count": 571.9890625, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9824184775352478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.867803167551756e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44634170532227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7518371582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19680, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.2994938433170319, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.012524006795138121, + "grad_norm": 0.0875559076666832, + "kept_token_count": 573.3296875, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9816354095935822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.173689300543629e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.57863235473633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5698486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19690, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.2998067229986191, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.012151578487828375, + "grad_norm": 0.09956879913806915, + "kept_token_count": 572.909375, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9824349880218506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.861353540443815e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.53255844116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.10581970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19700, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2989737629890442, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.010934750968590378, + "grad_norm": 0.0720425397157669, + "kept_token_count": 573.55, + "learning_rate": 8.65972705596018e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9822181165218353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.946065404918045e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.07635726928712, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.49587097167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19710, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.30043137073516846, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.012967792619019747, + "grad_norm": 0.0854981318116188, + "kept_token_count": 572.625, + "learning_rate": 8.64474589885725e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9835630357265472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.420707577490248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.84392395019532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.93678894042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19720, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2992514193058014, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.012023700354620814, + "grad_norm": 0.09230303764343262, + "kept_token_count": 573.190625, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9832396090030671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.547044176841155e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.9996726989746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3234344482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19730, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.29955483973026276, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.012729316391050816, + "grad_norm": 0.11070031672716141, + "kept_token_count": 572.9171875, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9827434718608856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.740848475601524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.19964218139648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5674591064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19740, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.3004457771778107, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.012810960551723838, + "grad_norm": 0.0698811337351799, + "kept_token_count": 572.421875, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.982038801908493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.016116142040118e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.32564010620118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.37181701660154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19750, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29929443895816804, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.014944521989673377, + "grad_norm": 0.09027647227048874, + "kept_token_count": 573.2828125, + "learning_rate": 8.584898674669772e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9815694868564606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.199435458460357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.15287475585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.06603088378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19760, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3004975259304047, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.011628117645159364, + "grad_norm": 0.09573561698198318, + "kept_token_count": 572.321875, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9831065833568573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.599013904633466e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.35033340454102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5877960205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19770, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.30016596913337706, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.011994770122691989, + "grad_norm": 0.09663888067007065, + "kept_token_count": 572.575, + "learning_rate": 8.555021759597627e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9831086456775665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.598203617613763e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.54814987182617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.94098510742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19780, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3006024181842804, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.014367709681391717, + "grad_norm": 0.07445655018091202, + "kept_token_count": 572.415625, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9824034869670868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.873653837828897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.6332405090332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9397399902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19790, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.29930207431316375, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.013923162082210182, + "grad_norm": 0.08057533949613571, + "kept_token_count": 573.4109375, + "learning_rate": 8.525176169541095e-06, + "loss": 0.014, + "mgd_cos_sim": 0.983380264043808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.492104294011369e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3382957458496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5911804199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.29978967607021334, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.01243295455351472, + "grad_norm": 0.07984716445207596, + "kept_token_count": 573.046875, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9834856986999512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.450914188462775e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.71820068359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6493377685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19810, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.29931150674819945, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.011150841927155852, + "grad_norm": 0.08852290362119675, + "kept_token_count": 573.284375, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9813669264316559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.278564225998707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.41793746948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.30654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19820, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3010332375764847, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.013561907224357128, + "grad_norm": 0.12238167971372604, + "kept_token_count": 571.953125, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9824051439762116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.873005149827804e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.26305084228515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8890441894531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19830, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2986396253108978, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013071337155997753, + "grad_norm": 0.11272604763507843, + "kept_token_count": 573.675, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9822192668914795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.945619534235448e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.94206008911132, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9001159667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19840, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.2994369328022003, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.01257624225690961, + "grad_norm": 0.10029224306344986, + "kept_token_count": 573.2640625, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.982060557603836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.007609165157191e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.98455810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8751220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3003997325897217, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.01265139002352953, + "grad_norm": 0.12110777944326401, + "kept_token_count": 572.434375, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9825761079788208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.806224446336273e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.4815773010254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.45322265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29984374940395353, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.012327524786815047, + "grad_norm": 0.07778242230415344, + "kept_token_count": 572.9171875, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9825161278247834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.829655831097625e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.45652389526367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.25318908691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19870, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3000398129224777, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.014121556002646685, + "grad_norm": 0.11183519661426544, + "kept_token_count": 572.9203125, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9819528341293335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.049689593259245e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.6392204284668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.19142150878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19880, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.299895253777504, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.012706995662301778, + "grad_norm": 0.09997458010911942, + "kept_token_count": 572.75625, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9830992043018341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.601902496186085e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.04074096679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5561553955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19890, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.3001954615116119, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.012155970418825745, + "grad_norm": 0.08709950745105743, + "kept_token_count": 572.3859375, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9827849805355072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.724642262270208e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.50208969116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.73369750976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19900, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.30041566491127014, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.015224415529519319, + "grad_norm": 0.09937698394060135, + "kept_token_count": 572.3234375, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.981119441986084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.375236455118284e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.86875076293946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4823944091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.29997906684875486, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.013039465714246034, + "grad_norm": 0.10748112201690674, + "kept_token_count": 572.6, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9817779183387756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.118018475011922e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57539138793945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.62779846191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19920, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.3018886923789978, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.017701972741633654, + "grad_norm": 0.11585593223571777, + "kept_token_count": 571.240625, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9828338623046875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.705543783027678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.99891204833985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5224578857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19930, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2991418123245239, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.012208923744037748, + "grad_norm": 0.07964326441287994, + "kept_token_count": 573.565625, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9824274718761444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.864294264232739e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07428741455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4292846679688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19940, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.3006716310977936, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.012920222897082567, + "grad_norm": 0.10451348125934601, + "kept_token_count": 572.0515625, + "learning_rate": 8.302347345847146e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9828377723693847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.704005354549736e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7791763305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.12255859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19950, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.3001197844743729, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.01562486649490893, + "grad_norm": 0.1263045072555542, + "kept_token_count": 572.678125, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9826289236545562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.785593577660621e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55501403808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.64092407226565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19960, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3002593368291855, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.011542405746877194, + "grad_norm": 0.1275874823331833, + "kept_token_count": 572.546875, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9835614264011383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.42133814835688e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.89713668823242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2096252441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19970, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.29844469726085665, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.012515003699809313, + "grad_norm": 0.08541937917470932, + "kept_token_count": 573.896875, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9813239574432373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.295341565622948e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.47816009521485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8891998291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19980, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.3007470369338989, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.011126113636419178, + "grad_norm": 0.0823843851685524, + "kept_token_count": 572.121875, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9823654115200042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.888533862365875e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.39468154907226, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.63270568847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19990, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29995906352996826, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.013377340324223042, + "grad_norm": 0.10108593851327896, + "kept_token_count": 572.65625, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.982460331916809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.851448670204263e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.53968124389648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5735168457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20000, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.2997604638338089, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.011041954159736633, + "grad_norm": 0.12537306547164917, + "kept_token_count": 572.95, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9826478362083435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.778209572075867e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.025048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.28697814941404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20010, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3006403535604477, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.012605216354131699, + "grad_norm": 0.10906583815813065, + "kept_token_count": 572.178125, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9820570111274719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.009007276792545e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93127746582032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.38844604492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20020, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3005900174379349, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.01427999408915639, + "grad_norm": 0.10694371908903122, + "kept_token_count": 572.2375, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9829444468021393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.662342966592404e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.25367813110351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.92406311035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20030, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2991517513990402, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.012608484458178281, + "grad_norm": 0.07089847326278687, + "kept_token_count": 573.4609375, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9829912662506104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.644053682975936e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.72371826171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5699768066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20040, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2994502365589142, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.010352531168609857, + "grad_norm": 0.08386800438165665, + "kept_token_count": 573.309375, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9824612438678741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.851093712612055e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7495132446289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16358947753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20050, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.30053969025611876, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.01149904252961278, + "grad_norm": 0.10186843574047089, + "kept_token_count": 572.303125, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.981564861536026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.201240659924225e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.78532180786132, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2130554199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20060, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2999526411294937, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.01294222129508853, + "grad_norm": 0.08736073970794678, + "kept_token_count": 572.7359375, + "learning_rate": 8.125399750837202e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9827607631683349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.734101116308011e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.28876724243165, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.72157592773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20070, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30071897208690646, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.013429951015859842, + "grad_norm": 0.12614239752292633, + "kept_token_count": 572.0796875, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9826331198215484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.783954813727178e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.69819488525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5645385742188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20080, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.3004112601280212, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.01237290888093412, + "grad_norm": 0.06250854581594467, + "kept_token_count": 572.54375, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9830317318439483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.628248920605983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.26400756835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9147674560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20090, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.3002659440040588, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.01216332702897489, + "grad_norm": 0.0893959328532219, + "kept_token_count": 572.6453125, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9828145921230316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.713072143611498e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.42275314331054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.13548583984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20100, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.30036321878433225, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.01418001581914723, + "grad_norm": 0.09981632977724075, + "kept_token_count": 572.490625, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9822375118732453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.938490951142739e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7904541015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4039581298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29976312816143036, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.012193222623318433, + "grad_norm": 0.10951347649097443, + "kept_token_count": 573.046875, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9825259208679199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.825830860179849e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57470397949218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.19959716796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30053756237030027, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.012860642746090889, + "grad_norm": 0.10541190207004547, + "kept_token_count": 572.225, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9829704225063324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.652193806075957e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5863052368164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.570654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20130, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3007537811994553, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.012515864241868258, + "grad_norm": 0.10240599513053894, + "kept_token_count": 572.1421875, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9821952879428864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.954988930374384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.3599349975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8402130126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20140, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.29968576729297636, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.012744074082002043, + "grad_norm": 0.0782790333032608, + "kept_token_count": 572.9171875, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9822315990924835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.940803505131043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.53329849243164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91656188964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20150, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.30050415694713595, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.013607325125485658, + "grad_norm": 0.0962529107928276, + "kept_token_count": 572.25, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9823875784873962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.879865395603701e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.53554153442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6196319580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20160, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30046737790107725, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.014116046344861388, + "grad_norm": 0.09492156654596329, + "kept_token_count": 572.659375, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.982667738199234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.770435647922568e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5578742980957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.52202758789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20170, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.29998838901519775, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.011349424440413713, + "grad_norm": 0.07631415873765945, + "kept_token_count": 572.740625, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9830177426338196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.633718003286048e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3345848083496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7599212646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20180, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.29950866997241976, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.01301164049655199, + "grad_norm": 0.0809599831700325, + "kept_token_count": 573.2671875, + "learning_rate": 7.949655011814968e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9814683377742768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.238946491270327e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.300341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.79624938964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20190, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2999241232872009, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.012256250251084566, + "grad_norm": 0.11856682598590851, + "kept_token_count": 572.6609375, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9828733801841736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.690103618893773e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.24890670776367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7368103027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20200, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2997554361820221, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.01651748428121209, + "grad_norm": 0.09955736994743347, + "kept_token_count": 572.9265625, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9818741738796234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.08041807229165e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3844108581543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0695465087891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20210, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.3010401725769043, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.013837875053286552, + "grad_norm": 0.09841366857290268, + "kept_token_count": 572.040625, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9823795616626739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.883000933157745e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.26453857421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3900146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20220, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.299907386302948, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.012859076634049416, + "grad_norm": 0.09883906692266464, + "kept_token_count": 572.8359375, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9818988561630249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.070776191540063e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.60801391601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.27960815429685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20230, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3008665472269058, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.012736180797219276, + "grad_norm": 0.12166283279657364, + "kept_token_count": 572.05, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9829550445079803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.658201964455656e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44233856201171, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.56842041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20240, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.3000044941902161, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.011800118442624807, + "grad_norm": 0.08167729526758194, + "kept_token_count": 572.5578125, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9827247262001038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.748174964741339e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.16828079223633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7334838867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20250, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30081315636634826, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.012633945420384407, + "grad_norm": 0.09301240742206573, + "kept_token_count": 572.065625, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9824012637138366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.874526261526625e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.9317726135254, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1971466064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20260, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3000837564468384, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.01261450182646513, + "grad_norm": 0.08069536834955215, + "kept_token_count": 572.7546875, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9830526411533356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.620086278417148e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.59208908081055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8159118652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20270, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.30002327263355255, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.01200866959989071, + "grad_norm": 0.09316455572843552, + "kept_token_count": 572.74375, + "learning_rate": 7.81865433804676e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9828580379486084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.696097261738032e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23288650512696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.75954284667966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20280, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2998863995075226, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.013469384098425508, + "grad_norm": 0.08188863098621368, + "kept_token_count": 572.8703125, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9831792652606964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.570624427695292e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.43596878051758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.82274169921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20290, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30032950043678286, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.01410954729653895, + "grad_norm": 0.07205860316753387, + "kept_token_count": 572.534375, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9819483101367951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.051463471725583e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5142807006836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1848663330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20300, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3002932548522949, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.01335033942013979, + "grad_norm": 0.12214081734418869, + "kept_token_count": 572.5953125, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9821203351020813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.984255014685914e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.12609786987305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.73773193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2994914472103119, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.011525776423513889, + "grad_norm": 0.08566999435424805, + "kept_token_count": 573.3046875, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9822230339050293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.944143897271715e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.7254867553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4933166503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.3004719078540802, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.01233886876143515, + "grad_norm": 0.08427082002162933, + "kept_token_count": 572.56875, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9829881012439727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.64528961351607e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.30926818847657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0373840332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20330, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2998018205165863, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.012810406368225813, + "grad_norm": 0.08596643060445786, + "kept_token_count": 573.1109375, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.982766580581665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.73182585160248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.98515396118164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7324493408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20340, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2996091842651367, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.012940498534590006, + "grad_norm": 0.11198826134204865, + "kept_token_count": 573.0765625, + "learning_rate": 7.717252920010828e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9820288121700287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.020014927547891e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.77825698852538, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6693176269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2997646629810333, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.012293926207348704, + "grad_norm": 0.0958174616098404, + "kept_token_count": 572.9546875, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9838981688022613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.289794073381927e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75541381835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4339111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20360, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30043081045150755, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.013436573557555676, + "grad_norm": 0.10259489715099335, + "kept_token_count": 572.134375, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9832705140113831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.534977474075277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9943473815918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9996368408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20370, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.2990052193403244, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.015282668359577655, + "grad_norm": 0.09189503639936447, + "kept_token_count": 573.61875, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9830119252204895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.635986901528667e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.31971817016601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2211181640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20380, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2995910316705704, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.011043394589796663, + "grad_norm": 0.0961611196398735, + "kept_token_count": 573.075, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9827544271945954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.73656839353498e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.52928466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6807037353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20390, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.30004982352256776, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.012975784810259938, + "grad_norm": 0.10370286554098129, + "kept_token_count": 572.6546875, + "learning_rate": 7.645088460126193e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9828880965709687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.684352338197641e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.10576095581055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2319854736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20400, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.29991625249385834, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.011153397476300597, + "grad_norm": 0.10050696134567261, + "kept_token_count": 572.81875, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9821777224540711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.961842882446945e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.41853637695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7822021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20410, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.29902231097221377, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.01229870468378067, + "grad_norm": 0.10772018879652023, + "kept_token_count": 573.6734375, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9826592981815339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.773733221052681e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.61008758544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.92139892578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20420, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.29955722391605377, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.010894149495288729, + "grad_norm": 0.07268907874822617, + "kept_token_count": 573.0609375, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9829431354999543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.662856831098906e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.89480361938476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.29444274902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20430, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.29859026372432707, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.0132234669290483, + "grad_norm": 0.11935736238956451, + "kept_token_count": 573.834375, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9828931450843811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.682380590063985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.7744384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4341064453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20440, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.2992697715759277, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.012448286078870297, + "grad_norm": 0.08292485773563385, + "kept_token_count": 573.4515625, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9827923834323883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.721749596181326e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.58513412475585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7526062011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20450, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2998927623033524, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.011439345264807343, + "grad_norm": 0.08946923166513443, + "kept_token_count": 572.6953125, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9824901223182678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.839811940153595e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.3973762512207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.19390258789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20460, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.29875310957431794, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.012244019377976655, + "grad_norm": 0.11029023677110672, + "kept_token_count": 573.58125, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9823559403419495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.892228411743417e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.93017349243163, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5217803955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20470, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.2993926078081131, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.015621377620846033, + "grad_norm": 0.09657146781682968, + "kept_token_count": 573.2640625, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.982921838760376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.67117492412217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.86623001098633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5459381103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20480, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2998763293027878, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.012563420087099075, + "grad_norm": 0.09972608834505081, + "kept_token_count": 572.9984375, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9823560893535614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.892171768413391e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.46188659667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6035491943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20490, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.300539568066597, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.010122340917587281, + "grad_norm": 0.09076286852359772, + "kept_token_count": 572.2296875, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9825398564338684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.820383568992839e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4541732788086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.53940734863284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20500, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2997003376483917, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.012853693123906851, + "grad_norm": 0.12225305289030075, + "kept_token_count": 572.8171875, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9826498627662659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.777419184800237e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8804214477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.69896545410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20510, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.30121009051799774, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.010758453141897916, + "grad_norm": 0.07652521133422852, + "kept_token_count": 571.78125, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9834551095962525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.46286214760039e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.14182357788086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.96868896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3001921236515045, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013380340952426194, + "grad_norm": 0.08626837283372879, + "kept_token_count": 572.5515625, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9824267983436584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.864555689389817e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.03768005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7855529785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20530, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.300964492559433, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.013078991696238518, + "grad_norm": 0.09267468750476837, + "kept_token_count": 571.9765625, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9822332620620727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.940149469301105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.8590476989746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.38988342285154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20540, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3001288026571274, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.012423845985904337, + "grad_norm": 0.16766583919525146, + "kept_token_count": 572.9828125, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9831243336200715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.592070858459919e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90180740356445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.36528930664065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20550, + "valid_token_count": 818.70625 + }, + { + "actual_token_mask_ratio": 0.30040199458599093, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.01151714138686657, + "grad_norm": 0.0913812667131424, + "kept_token_count": 572.434375, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9825663805007935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.810026243329049e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.42759704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.01963195800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20560, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29931787252426145, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.01316366232931614, + "grad_norm": 0.11974109709262848, + "kept_token_count": 573.44375, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9825122892856598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831161445006729e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.31364517211914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3417083740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20570, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.2996080696582794, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.013973947521299124, + "grad_norm": 0.10102704912424088, + "kept_token_count": 573.00625, + "learning_rate": 7.387162003176037e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9822442293167114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.9358686596388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.53788375854492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9591125488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20580, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3001021981239319, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.011850684089586139, + "grad_norm": 0.0792272686958313, + "kept_token_count": 572.4640625, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9836477875709534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.38760204310529e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.13949890136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.78602905273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20590, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.2999601811170578, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.012591596320271492, + "grad_norm": 0.11943137645721436, + "kept_token_count": 572.7203125, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9827202081680297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.749936801497824e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93282775878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.11561584472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20600, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2984468400478363, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.012756261602044105, + "grad_norm": 0.08483660966157913, + "kept_token_count": 574.1515625, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9820305585861206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.019329132162966e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.43026351928711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1093505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20610, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.3005112260580063, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.012162981554865837, + "grad_norm": 0.09248405694961548, + "kept_token_count": 572.4671875, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9826403021812439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.781151714676525e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.14023513793946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.69493408203124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20620, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2994213253259659, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.013607564009726047, + "grad_norm": 0.0876067578792572, + "kept_token_count": 573.2234375, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9819935858249664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.033771253190934e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.10703735351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3928497314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20630, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3014209657907486, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.012519546225667, + "grad_norm": 0.13900388777256012, + "kept_token_count": 571.509375, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.982820063829422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.710931120323949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.09515075683593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7082092285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.3000175029039383, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.011204557726159692, + "grad_norm": 0.08139139413833618, + "kept_token_count": 572.6859375, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9829907774925232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.644248715019785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.44220123291015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.00490112304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20650, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30012000501155855, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.012573990505188704, + "grad_norm": 0.10647137463092804, + "kept_token_count": 572.9390625, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9817387163639069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.133332983357831e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.02694168090821, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.85613708496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20660, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.299335914850235, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.011081043351441621, + "grad_norm": 0.0666295662522316, + "kept_token_count": 573.565625, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9830681562423706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614020130655263e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.64514846801758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.42888793945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20670, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.3000713914632797, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.013560399087145925, + "grad_norm": 0.09227343648672104, + "kept_token_count": 572.484375, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9823446631431579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.896636477904394e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.25653305053712, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0531524658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20680, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.30048397481441497, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.01312726130709052, + "grad_norm": 0.10602044314146042, + "kept_token_count": 572.60625, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9819954037666321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.033057190710678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.75279235839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0502044677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20690, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.2996022909879684, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014309339504688979, + "grad_norm": 0.0928882360458374, + "kept_token_count": 573.15, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9812759518623352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.314097238122485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.67514877319336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.707568359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.30026518404483793, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.01183345690369606, + "grad_norm": 0.1369915008544922, + "kept_token_count": 572.6125, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9819253444671631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.060430107230786e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.57714309692383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.06305541992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20710, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.30135825276374817, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.014687434118241072, + "grad_norm": 0.09995682537555695, + "kept_token_count": 571.7328125, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9821325480937958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.979492682148702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.05578231811523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.67578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20720, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30003391206264496, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.012819333001971245, + "grad_norm": 0.07488340139389038, + "kept_token_count": 572.9078125, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9832044541835785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.560782858286985e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2427146911621, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1619537353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20730, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.3001353293657303, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.01141766314394772, + "grad_norm": 0.10176856070756912, + "kept_token_count": 572.6796875, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.98379185795784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.331323565973434e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5603141784668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4779479980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20740, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.29991512298583983, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.011996844597160817, + "grad_norm": 0.09640002995729446, + "kept_token_count": 573.0203125, + "learning_rate": 7.146314960931996e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9832200527191162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.554689098265953e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.4479751586914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.53785095214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20750, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2993084639310837, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.010375429457053543, + "grad_norm": 0.10102088749408722, + "kept_token_count": 573.290625, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9824298024177551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.863378512207418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.46500091552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6351806640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20760, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.2999851733446121, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.012648356147110462, + "grad_norm": 0.09390031546354294, + "kept_token_count": 572.859375, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9826178789138794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.789908584323712e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07201232910157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5659942626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20770, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.3003907442092896, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.012786195520311594, + "grad_norm": 0.0963912308216095, + "kept_token_count": 572.7453125, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.982799643278122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.718911936332006e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.62084503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.90399780273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20780, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.2994565933942795, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.012061627116054296, + "grad_norm": 0.08886346220970154, + "kept_token_count": 573.0734375, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9831228852272034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.592637619178277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.05497665405274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3236572265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20790, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.2998567223548889, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.010530050937086343, + "grad_norm": 0.08125060051679611, + "kept_token_count": 572.7734375, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.982490473985672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.839675079390872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.96035919189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8933563232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20800, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.3002079904079437, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.014017881918698549, + "grad_norm": 0.10309851914644241, + "kept_token_count": 572.7421875, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9825958132743835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.798528193030506e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.81071090698242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.46600036621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20810, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.3005578935146332, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.012562335282564164, + "grad_norm": 0.09767458587884903, + "kept_token_count": 572.1765625, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9822175204753876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.946302200958599e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.93884963989258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.80914611816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20820, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.29985072910785676, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.014546309877187014, + "grad_norm": 0.10841281712055206, + "kept_token_count": 573.025, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9819159030914306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.064116070978343e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.07688827514649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.3006405770778656, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.011248430609703064, + "grad_norm": 0.10707499086856842, + "kept_token_count": 572.25625, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9822461545467377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.935114906809758e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.41408004760743, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.89116821289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20840, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30049634873867037, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.012724715191870928, + "grad_norm": 0.11885934323072433, + "kept_token_count": 572.3515625, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9828463792800903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.700658050249331e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20418014526368, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6585754394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20850, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.300269877910614, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.012407776899635792, + "grad_norm": 0.13460877537727356, + "kept_token_count": 572.4390625, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9832330763339996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.54959883831907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.80442276000977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48939819335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20860, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2994811564683914, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.012575451750308275, + "grad_norm": 0.08822420239448547, + "kept_token_count": 573.23125, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9827574789524078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.735375645803288e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.38478775024414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4140930175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20870, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29920504689216615, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.011020713904872536, + "grad_norm": 0.08314748108386993, + "kept_token_count": 573.3671875, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9835758328437805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.41571055894019e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57306213378907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.79393310546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20880, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.29906754195690155, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.013155394513159991, + "grad_norm": 0.14539264142513275, + "kept_token_count": 573.6453125, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9817733824253082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.119793044694234e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83584060668946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4430816650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20890, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2999752789735794, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.013887951150536537, + "grad_norm": 0.08646783232688904, + "kept_token_count": 573.0203125, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9822314262390137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.940865423530341e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.19936447143554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.69373779296876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20900, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.30068414509296415, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.01337651452049613, + "grad_norm": 0.08573688566684723, + "kept_token_count": 572.2421875, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.981594443321228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.189685056800954e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8378303527832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4094573974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20910, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.29963076710700987, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.013757351133972407, + "grad_norm": 0.08204201608896255, + "kept_token_count": 573.1234375, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9834503591060638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.464724283432588e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.31069564819336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.09039611816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20920, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29973676800727844, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.01181134171783924, + "grad_norm": 0.35423967242240906, + "kept_token_count": 573.2671875, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9823887407779693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.87941887008492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52129287719727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3586059570313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20930, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.300323149561882, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.013163668243214488, + "grad_norm": 0.11613168567419052, + "kept_token_count": 572.5140625, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9820443868637085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.01392626069719e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.1621696472168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.52716674804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20940, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29917094111442566, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.01266665980219841, + "grad_norm": 0.0773807242512703, + "kept_token_count": 573.334375, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9830665946006775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614629746763967e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42899856567382, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.07041320800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.2998710423707962, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.013484249264001847, + "grad_norm": 0.09063641726970673, + "kept_token_count": 572.8671875, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9824077129364014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.872007907077204e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2135726928711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.32764587402346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20960, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.3000605881214142, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.013735135830938817, + "grad_norm": 0.17672099173069, + "kept_token_count": 572.784375, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9824112474918365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.870628058095463e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75099716186523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.43812866210936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20970, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29998103678226473, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.013491383567452431, + "grad_norm": 0.12179989367723465, + "kept_token_count": 572.96875, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9823820769786835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.882016132294665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.73698120117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3475830078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20980, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.29931910037994386, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.012673660879954696, + "grad_norm": 0.09618271142244339, + "kept_token_count": 573.2046875, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9829775750637054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.64940154820215e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9133804321289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8575500488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20990, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.3000063180923462, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.011031364789232612, + "grad_norm": 0.05970054864883423, + "kept_token_count": 572.71875, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9821538329124451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.971182156121359e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.00991897583008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0160675048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21000, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.30012614727020265, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.015224511176347733, + "grad_norm": 0.10202169418334961, + "kept_token_count": 572.7125, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9833888530731201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.488746766990517e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.07256927490235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3947723388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.3002351552248001, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.013969704788178206, + "grad_norm": 0.09710156172513962, + "kept_token_count": 572.365625, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9818804681301116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.077959380694666e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.54721908569336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.51646728515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21020, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.30109187960624695, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.011957634938880801, + "grad_norm": 0.10261917114257812, + "kept_token_count": 572.0578125, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9822733461856842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.924493573023938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.57393264770508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2494201660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21030, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.3003721863031387, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.013711364194750785, + "grad_norm": 0.08643326908349991, + "kept_token_count": 572.459375, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.982933086156845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.666785448032897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38469085693359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8685607910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21040, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30020729303359983, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.012724051624536515, + "grad_norm": 0.07352056354284286, + "kept_token_count": 572.6359375, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.982650351524353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.77722055115737e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.05267333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.82896728515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21050, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30002333521842955, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.01141473618336022, + "grad_norm": 0.0876159593462944, + "kept_token_count": 572.7578125, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9834238648414612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.475071022578049e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.94924087524414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6140655517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21060, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30078351497650146, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.01036079335026443, + "grad_norm": 0.07881885021924973, + "kept_token_count": 572.1703125, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9822582006454468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.93040681653656e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.86830825805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21070, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30058842301368716, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.014716226421296597, + "grad_norm": 0.10156778991222382, + "kept_token_count": 572.3875, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.980943089723587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.444125803885981e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.37023391723633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.02988891601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21080, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3001533716917038, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.011768284626305104, + "grad_norm": 0.09423321485519409, + "kept_token_count": 572.6515625, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9829046726226807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.677879864582791e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.56535110473632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6302520751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21090, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.30014788508415224, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.012370018614456057, + "grad_norm": 0.1172836571931839, + "kept_token_count": 572.71875, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9829068183898926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.677039818896447e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61399612426757, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6867614746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21100, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.300109925866127, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.011949497973546386, + "grad_norm": 0.08939502388238907, + "kept_token_count": 572.590625, + "learning_rate": 6.645492438179188e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9829080879688263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.67655433062464e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.236083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.98138732910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21110, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.3010177046060562, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.012604286475107074, + "grad_norm": 0.09329438954591751, + "kept_token_count": 572.0671875, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9822446525096893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.935697456356138e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.84906768798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3721496582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21120, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.29965011179447176, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.013600739184767008, + "grad_norm": 0.1233852356672287, + "kept_token_count": 573.184375, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.981879609823227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.0782953480375e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.44444580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2599243164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21130, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30072844922542574, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.01167610865086317, + "grad_norm": 0.09553435444831848, + "kept_token_count": 571.9859375, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9827306210994721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.745870050508528e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.24053573608398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9528289794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21140, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.3009689062833786, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.01205388391390443, + "grad_norm": 0.07868831604719162, + "kept_token_count": 571.85625, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9833561658859253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.501514581032097e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42043609619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0633178710938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21150, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.29813616871833803, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.010714920470491052, + "grad_norm": 0.08649049699306488, + "kept_token_count": 574.18125, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.983219587802887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.554869360115844e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.08320617675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.04396057128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.30090391933918, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.012546656280755996, + "grad_norm": 0.09169825166463852, + "kept_token_count": 572.2953125, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.982562917470932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.811379280406982e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.15626602172851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.29444580078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21170, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.3007213622331619, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.012433110224083065, + "grad_norm": 0.09348201006650925, + "kept_token_count": 572.303125, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9829374194145203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.665091095783282e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.3818145751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.797021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.3008826971054077, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.014111870108172298, + "grad_norm": 0.12814071774482727, + "kept_token_count": 571.934375, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9819113194942475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.06591090420261e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38704452514648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6031860351562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21190, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2993637531995773, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.011830696044489742, + "grad_norm": 0.07644343376159668, + "kept_token_count": 573.4578125, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9826172113418579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.790166153223254e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38157958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.23846740722655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21200, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.301240062713623, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.012595942709594965, + "grad_norm": 0.09574593603610992, + "kept_token_count": 571.653125, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9818380296230316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.094540269463323e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.05790252685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2956939697266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21210, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.2999396830797195, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.01061548045836389, + "grad_norm": 0.10249125957489014, + "kept_token_count": 572.8359375, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9826470136642456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.778532551834359e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.44230270385742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5882232666016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.300117039680481, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.013093669293448329, + "grad_norm": 0.09622744470834732, + "kept_token_count": 572.4921875, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.983098816871643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.602045468753203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.94375991821289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0559356689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21230, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.29960650503635405, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.012542372988536953, + "grad_norm": 0.12537430226802826, + "kept_token_count": 573.209375, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.983280211687088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.531182552862446e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.09589004516602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.30824279785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21240, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2992231070995331, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.010204737726598977, + "grad_norm": 0.10548131167888641, + "kept_token_count": 573.41875, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9829882442951202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.645234534516931e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.60728225708007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41957702636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21250, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29982687532901764, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.010690133599564434, + "grad_norm": 0.09082984179258347, + "kept_token_count": 572.8703125, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9820457398891449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.013399954303168e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.73062744140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9261505126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21260, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.30022911429405214, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.01178728574886918, + "grad_norm": 0.09449257701635361, + "kept_token_count": 572.4265625, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9827384769916534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.7428064357955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.21032638549805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7189422607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21270, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.2996063619852066, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.011613345379009843, + "grad_norm": 0.11886405199766159, + "kept_token_count": 573.03125, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9830798923969268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.609427255170886e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.40191116333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21280, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29994374215602876, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.011719530168920755, + "grad_norm": 0.06818440556526184, + "kept_token_count": 572.946875, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9822018146514893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.952434050617739e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.20262069702149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.95924682617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21290, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.30012943148612975, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.012248115427792073, + "grad_norm": 0.11442723870277405, + "kept_token_count": 572.721875, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.982207304239273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.950292918190825e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.46965942382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8373260498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21300, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3012323439121246, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.01427240315824747, + "grad_norm": 0.09080418199300766, + "kept_token_count": 571.8, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9832777202129364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.532163934025448e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39330596923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.387158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21310, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.2997050374746323, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.01089613800868392, + "grad_norm": 0.0962543934583664, + "kept_token_count": 573.0421875, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9823866009712219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.880255969008431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77467498779296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8554321289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21320, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.29956036508083345, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.011826421972364188, + "grad_norm": 0.08505015075206757, + "kept_token_count": 573.215625, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9823043584823609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.912382013979368e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.66463775634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.021435546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21330, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3008533209562302, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.012351051345467568, + "grad_norm": 0.10873769223690033, + "kept_token_count": 571.8625, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9823670148849487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.887902491143904e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.46275253295899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.64097595214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21340, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.2985160738229752, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.014215542003512383, + "grad_norm": 0.110869862139225, + "kept_token_count": 573.98125, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9819146275520325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.064626188366674e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.831884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.57591247558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.300667878985405, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.012549901800230145, + "grad_norm": 0.1387898027896881, + "kept_token_count": 572.1140625, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9830779731273651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.610191994695924e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2372657775879, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6354156494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21360, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29898851215839384, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.011014330992475152, + "grad_norm": 0.11594229936599731, + "kept_token_count": 573.546875, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9815727174282074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.198175153462216e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.65915756225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8627593994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21370, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3000460773706436, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.011147522367537022, + "grad_norm": 0.11335588246583939, + "kept_token_count": 572.784375, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9817700624465943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.121082235244103e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76926498413086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9641571044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21380, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.30136507749557495, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.01129587679170072, + "grad_norm": 0.10220666974782944, + "kept_token_count": 571.5625, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9822263836860656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.942842519492842e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83915328979492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.09613647460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.29996722340583803, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.012970052100718022, + "grad_norm": 0.10661706328392029, + "kept_token_count": 572.75625, + "learning_rate": 6.251582125592394e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9808441638946533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.482766159228049e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.65802230834962, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.06471252441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21400, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29962784349918364, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.012227210868149995, + "grad_norm": 0.07384377717971802, + "kept_token_count": 572.9453125, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9824790179729461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.844154522696044e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7729866027832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4657958984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.3011961579322815, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.01266189906746149, + "grad_norm": 0.10811974108219147, + "kept_token_count": 571.64375, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9824299454689026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.863320522825234e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.91390075683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4021850585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21420, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.2991896152496338, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.012137234210968018, + "grad_norm": 0.10691796243190765, + "kept_token_count": 573.4953125, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9818536520004273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.08843297616113e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.3814796447754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.85064392089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21430, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3002601355314255, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.012389043252915144, + "grad_norm": 0.11478505283594131, + "kept_token_count": 572.5359375, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9827417969703675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.741507349943277e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.95541763305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.91083068847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21440, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30103370547294617, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.010872862488031387, + "grad_norm": 0.08148235827684402, + "kept_token_count": 572.0484375, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9834639847278595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.459399955929256e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.93113555908204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.32960815429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21450, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.300115504860878, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.012076925905421376, + "grad_norm": 0.11476707458496094, + "kept_token_count": 572.6953125, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9829236090183258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.670486545772292e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61266174316407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.81354064941405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21460, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.3006646126508713, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.013245530426502228, + "grad_norm": 0.08658529072999954, + "kept_token_count": 572.29375, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9820092976093292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.027634128462524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.09842376708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.36083068847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21470, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29977831840515134, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.012602087808772922, + "grad_norm": 0.0906471535563469, + "kept_token_count": 572.9359375, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9823038339614868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.912581811775453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.08080139160157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5719757080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21480, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.2999024271965027, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.012464428413659334, + "grad_norm": 0.0886816531419754, + "kept_token_count": 572.675, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9832321882247925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.54994753858773e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.45785446166992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6501708984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21490, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.30123162269592285, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.011267211940139531, + "grad_norm": 0.0724417120218277, + "kept_token_count": 571.7203125, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9829000413417817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.679694779450074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83884353637696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7699401855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21500, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2991339176893234, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.012402424635365606, + "grad_norm": 0.11791913956403732, + "kept_token_count": 573.5359375, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.982325303554535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.904202964506113e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.78056030273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1776489257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21510, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.30121508836746214, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.013499329145997763, + "grad_norm": 0.11365094035863876, + "kept_token_count": 571.8515625, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9823407053947448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.898176143295131e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4596939086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.97882080078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21520, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.3001989841461182, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011268124915659428, + "grad_norm": 0.12698085606098175, + "kept_token_count": 572.8375, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9836049854755402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.404310588550288e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.17607116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.98412170410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21530, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.30042760968208315, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.011879268847405911, + "grad_norm": 0.1310214400291443, + "kept_token_count": 572.509375, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.981840831041336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.093434978742153e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.37515258789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.33839416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21540, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.30090897977352143, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.012428092351183296, + "grad_norm": 0.0930478572845459, + "kept_token_count": 571.9671875, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9826735913753509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.768152234144509e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.89699020385743, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.24803466796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21550, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30015788078308103, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.012072121724486351, + "grad_norm": 0.0885273888707161, + "kept_token_count": 572.803125, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9824227750301361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.86612580466317e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.72926330566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6831115722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21560, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.29885916113853456, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.012153065763413906, + "grad_norm": 0.10776619613170624, + "kept_token_count": 573.7796875, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9822478890419006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.934437515155878e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.05160369873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.20196838378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21570, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.30055008828639984, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.012429095059633254, + "grad_norm": 0.10211778432130814, + "kept_token_count": 572.3796875, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9830274701118469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.62991438730387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.60820007324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.05776062011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21580, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.30026315450668334, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.011562628438696265, + "grad_norm": 0.10334771871566772, + "kept_token_count": 572.6390625, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9831013321876526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.601055174542126e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.09480133056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.25909118652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21590, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.300595822930336, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.012114891968667506, + "grad_norm": 0.16793012619018555, + "kept_token_count": 572.38125, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9821104168891907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.98813732014969e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.06003341674804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.97069091796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21600, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.30080255270004275, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.012820289935916662, + "grad_norm": 0.09758433699607849, + "kept_token_count": 571.8921875, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.98239706158638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.876164407003671e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77039489746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.04613037109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21610, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.3010554939508438, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.012285568658262492, + "grad_norm": 0.10085628926753998, + "kept_token_count": 572.0078125, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9829424083232879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.663137901341543e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.32869567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0919708251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21620, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.3004766583442688, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.011993971420451999, + "grad_norm": 0.09239695966243744, + "kept_token_count": 572.5421875, + "learning_rate": 5.945507360378608e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9831072270870209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.598762702196837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.70830612182617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9363739013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21630, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.3000736892223358, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.01108848019503057, + "grad_norm": 0.08594395220279694, + "kept_token_count": 572.58125, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9828598022460937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.695405099890194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.95994491577149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.589599609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21640, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.2999584496021271, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.0124533720780164, + "grad_norm": 0.07320575416088104, + "kept_token_count": 573.053125, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9824981093406677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.836693355580792e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.36538238525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3162322998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21650, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.29941903948783877, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.010453237500041724, + "grad_norm": 0.06539636850357056, + "kept_token_count": 573.2921875, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9830829620361328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.60823530779453e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.76570205688476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5431701660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21660, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.30038578510284425, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.012541801575571299, + "grad_norm": 0.09331958740949631, + "kept_token_count": 572.1640625, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9827268242835998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.74735212669475e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.10532836914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.14130249023435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21670, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.29985156655311584, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.013216654956340789, + "grad_norm": 0.09651364386081696, + "kept_token_count": 572.9921875, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9822896122932434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.91813271259889e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82689056396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1912567138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21680, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.29931296706199645, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.010319013334810733, + "grad_norm": 0.1238802969455719, + "kept_token_count": 573.484375, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9831763565540313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.571748999704142e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.58604507446289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7955047607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21690, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.2990064650774002, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.013333881134167313, + "grad_norm": 0.08693584054708481, + "kept_token_count": 573.69375, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.982351666688919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.89389678882435e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5482078552246, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.352783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21700, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30054865777492523, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.011868189927190543, + "grad_norm": 0.08218662440776825, + "kept_token_count": 572.30625, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9821159482002259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.985970758250914e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.97015914916992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.77809448242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2997800916433334, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.012839395320042968, + "grad_norm": 0.08448877185583115, + "kept_token_count": 573.0109375, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9837662577629089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.341325606626925e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.85514831542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.98773498535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21720, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.29993548691272737, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.012175439484417438, + "grad_norm": 0.11225052177906036, + "kept_token_count": 572.5953125, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9828218758106232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.710222078254446e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.90407028198243, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3153442382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21730, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.30046993792057036, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.010394134325906634, + "grad_norm": 0.06687436997890472, + "kept_token_count": 572.390625, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.983810943365097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.323873094515875e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.2882293701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1201904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21740, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29986098110675813, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.013899623043835163, + "grad_norm": 0.14897520840168, + "kept_token_count": 572.7203125, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9836489617824554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.387142420862801e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.39572143554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2671264648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21750, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.3003766596317291, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.01214801510795951, + "grad_norm": 0.09296586364507675, + "kept_token_count": 572.578125, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9826652646064759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.771403532184195e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.46582412719727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3924255371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21760, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2997630417346954, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.011746213864535093, + "grad_norm": 0.12962645292282104, + "kept_token_count": 572.9828125, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9832418024539947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.546185722982046e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.15557403564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5600616455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21770, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29973122775554656, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.013232262525707483, + "grad_norm": 0.10024229437112808, + "kept_token_count": 572.98125, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9824794948101043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.843968731118366e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86276092529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.78709716796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21780, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3010156869888306, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.009610697161406279, + "grad_norm": 0.07479774951934814, + "kept_token_count": 571.9953125, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0096, + "mgd_cos_sim": 0.9829426050186157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.663061976723839e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21836395263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1792266845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29980946481227877, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.01205153251066804, + "grad_norm": 0.08745431154966354, + "kept_token_count": 572.865625, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.982252436876297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.932660689926707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.95641632080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7309600830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21800, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.29976252019405364, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.013458869932219387, + "grad_norm": 0.11341718584299088, + "kept_token_count": 573.0359375, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9825820028781891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.803931610193104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.33357391357421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.662109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2995574802160263, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.01231300923973322, + "grad_norm": 0.08992207050323486, + "kept_token_count": 573.2734375, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9829161822795868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.673382813460194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.6028564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.30763549804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21820, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.29995222091674806, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.012957157008349896, + "grad_norm": 0.08491034805774689, + "kept_token_count": 572.6546875, + "learning_rate": 5.68407916590721e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9838964462280273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.290469391387888e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.54392776489257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39446716308595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21830, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.29984464347362516, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.011712295562028885, + "grad_norm": 0.13233819603919983, + "kept_token_count": 572.7640625, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9826762676239014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.767101658624596e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.49436874389649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7337646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21840, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.30114070177078245, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.01319760880433023, + "grad_norm": 0.08404994010925293, + "kept_token_count": 571.9796875, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9830030500888824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.639451094088144e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.33563079833985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6858673095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21850, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.30017815828323363, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.01272480129264295, + "grad_norm": 0.09898320585489273, + "kept_token_count": 572.5578125, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.982860404253006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.695173215121031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.50405807495117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3549377441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21860, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.29985179007053375, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.012759311217814684, + "grad_norm": 0.0912410244345665, + "kept_token_count": 573.015625, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9828965127468109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.681070808554069e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.99542236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5083526611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21870, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.3000393509864807, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.012832044437527657, + "grad_norm": 0.11142001301050186, + "kept_token_count": 572.7609375, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9830598115921021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.617280232603661e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07096099853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.12489318847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2993649303913116, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.012575050489977002, + "grad_norm": 0.12566231191158295, + "kept_token_count": 573.54375, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9835675179958343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.418954035325442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.96289749145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2327880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21890, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.30048205256462096, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.011086628446355462, + "grad_norm": 0.05483074486255646, + "kept_token_count": 572.453125, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9827173590660095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.751045439159497e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.4565643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4283721923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21900, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29931565225124357, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.012141906935721637, + "grad_norm": 0.11867422610521317, + "kept_token_count": 573.3484375, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9835457324981689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.427462103602011e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.8867401123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2409912109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21910, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30061097741127013, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.011436832416802645, + "grad_norm": 0.1282881498336792, + "kept_token_count": 572.28125, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9828599214553833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.695356350974179e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.87836074829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.450439453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.30006733536720276, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.014231899566948415, + "grad_norm": 0.09953838586807251, + "kept_token_count": 572.821875, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9830095708370209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.636905309278517e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.53668746948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.28119506835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21930, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.30008744597435, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.013034201320260763, + "grad_norm": 0.13011591136455536, + "kept_token_count": 572.7828125, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9821925580501556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.956050274311564e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.92798461914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.32472534179686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21940, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30031818151474, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.012310890760272742, + "grad_norm": 0.10771017521619797, + "kept_token_count": 572.7203125, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.982627010345459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.78634169162251e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27867431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2751037597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21950, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2999993622303009, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.011583548784255982, + "grad_norm": 0.09828820824623108, + "kept_token_count": 572.8296875, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9834689021110534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.45747328235302e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.54420166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.26784362792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21960, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.3000874727964401, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.011963353678584098, + "grad_norm": 0.07978466153144836, + "kept_token_count": 572.7890625, + "learning_rate": 5.503765942568484e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9834609627723694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.460585645982065e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.80034408569335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4265991210938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21970, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29959011673927305, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.013286105543375015, + "grad_norm": 0.08206242322921753, + "kept_token_count": 573.09375, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9819035828113556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.06893675669562e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.16776123046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.66741943359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3003400981426239, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.012055858457460999, + "grad_norm": 0.09245169907808304, + "kept_token_count": 572.3046875, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9822855234146118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.919735897099599e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42027969360352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62127075195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21990, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.299678573012352, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.012743673333898187, + "grad_norm": 0.11204706877470016, + "kept_token_count": 572.990625, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9832376658916473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.547807133756578e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07850189208985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0358123779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22000, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.30028098821640015, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.011722473613917827, + "grad_norm": 0.12105847150087357, + "kept_token_count": 572.5640625, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9818698823451996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.08209459844511e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.10463790893554, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6518280029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22010, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3001670867204666, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.012628581933677196, + "grad_norm": 0.12464811652898788, + "kept_token_count": 572.659375, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9831610739231109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.577727399417199e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.56115188598633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5242126464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22020, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3006756603717804, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.011533333733677865, + "grad_norm": 0.0949048325419426, + "kept_token_count": 572.090625, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.983090603351593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.605255366594065e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.06891555786133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.254736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22030, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29981225430965425, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.011915519833564758, + "grad_norm": 0.16462543606758118, + "kept_token_count": 572.928125, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9830222308635712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.631953801843338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.84499893188476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.52186279296876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30049183070659635, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.012224029190838337, + "grad_norm": 0.0791129320859909, + "kept_token_count": 572.365625, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.983106255531311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.599135813303291e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.78289260864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4799346923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3012815684080124, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.00915216482244432, + "grad_norm": 0.09447597712278366, + "kept_token_count": 571.60625, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9828523218631744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.69833523716079e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57004241943359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2914978027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22060, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.3002941578626633, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.011775880958884955, + "grad_norm": 0.07205377519130707, + "kept_token_count": 572.375, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9832330167293548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.549622921738773e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.75378875732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.77293395996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22070, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29899721741676333, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.012346695130690932, + "grad_norm": 0.08499392867088318, + "kept_token_count": 573.7203125, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.982941061258316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.663665408268571e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.8571533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.82542724609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22080, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3011679857969284, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.010911317309364676, + "grad_norm": 0.12148701399564743, + "kept_token_count": 571.8328125, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9837141275405884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.361683590512257e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47665557861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2482574462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22090, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.300283345580101, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.011718560010194778, + "grad_norm": 0.06499514728784561, + "kept_token_count": 572.7296875, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9833549499511719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.501983989437576e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.57110214233398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8730163574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22100, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.30102076232433317, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.013162622461095452, + "grad_norm": 0.12203001230955124, + "kept_token_count": 572.178125, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9829336524009704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66655814711703e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.66661224365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.61700439453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22110, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.30009070932865145, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.010085434932261706, + "grad_norm": 0.09681430459022522, + "kept_token_count": 572.7234375, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9827447235584259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.740363533026539e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22639846801758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3073974609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22120, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.2997377783060074, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.013166421046480536, + "grad_norm": 0.07733682543039322, + "kept_token_count": 573.3421875, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9826573729515076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.774480643798598e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.11581649780274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91761474609376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22130, + "valid_token_count": 818.75625 + }, + { + "actual_token_mask_ratio": 0.29858354926109315, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.011677191639319063, + "grad_norm": 0.10191324353218079, + "kept_token_count": 573.8625, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9830067873001098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.637994556513149e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.26263122558593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.87437438964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22140, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.30074872374534606, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.011482675559818745, + "grad_norm": 0.10329552739858627, + "kept_token_count": 571.99375, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9826610684394836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.773042732675095e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.59691543579102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5010192871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22150, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.3007960975170135, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.014727010577917098, + "grad_norm": 0.06820916384458542, + "kept_token_count": 572.2578125, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9828216135501862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.710328452754765e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36153259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5161590576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22160, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.3008416503667831, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.011973660020157696, + "grad_norm": 0.10556905716657639, + "kept_token_count": 572.1109375, + "learning_rate": 5.250109272299523e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9834222793579102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.475692935055122e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.36592559814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5731536865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22170, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.3001240223646164, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.01169793219305575, + "grad_norm": 0.09020712971687317, + "kept_token_count": 572.53125, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9822018504142761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.95242422807496e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.04183197021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4203247070312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22180, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.300213822722435, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.010707653313875198, + "grad_norm": 0.08335844427347183, + "kept_token_count": 572.4671875, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9821692407131195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.965158900129609e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61544342041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6615020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.29940001368522645, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.01215064930729568, + "grad_norm": 0.10475558787584305, + "kept_token_count": 573.2265625, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9822259843349457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.942988002265338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.04651107788087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.24371948242185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22200, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2993124514818192, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.0143773652613163, + "grad_norm": 0.10474411398172379, + "kept_token_count": 573.3796875, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9825368404388428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.8215663486626e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.46154327392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66883850097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22210, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3000497162342072, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.01037006969563663, + "grad_norm": 0.08710785210132599, + "kept_token_count": 572.75, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.983659428358078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.383053441822994e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.22689514160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.25304260253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22220, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29866993725299834, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.011022567935287952, + "grad_norm": 0.08796237409114838, + "kept_token_count": 573.721875, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9835716962814331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.417331533157267e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.25569000244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.13251342773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.300038680434227, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.010764240426942706, + "grad_norm": 0.09476099908351898, + "kept_token_count": 572.8609375, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9825089514255524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.832453254901338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.82977294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0638732910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22240, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.3003951162099838, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.011171194445341825, + "grad_norm": 0.07637351751327515, + "kept_token_count": 572.346875, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9836260437965393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.396091048372909e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.526220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.82911071777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22250, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2998995393514633, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.010990770533680916, + "grad_norm": 0.13000871241092682, + "kept_token_count": 572.9703125, + "learning_rate": 5.137502456402705e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9830529451370239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.61996007693233e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72118911743163, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4084991455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22260, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2997203886508942, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.012445210292935372, + "grad_norm": 0.09074672311544418, + "kept_token_count": 572.9234375, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.982850843667984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.698904180666432e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11316757202148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.45844116210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22270, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2998466491699219, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.01318446029908955, + "grad_norm": 0.11762488633394241, + "kept_token_count": 572.7, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9820072412490845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.028442123555579e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.93795471191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2505584716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22280, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.29936121106147767, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.010740564856678247, + "grad_norm": 0.07698497921228409, + "kept_token_count": 573.3703125, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9825511455535889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.815975575591438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.85689392089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2355987548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22290, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.29920826852321625, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.013521816581487656, + "grad_norm": 0.0952039361000061, + "kept_token_count": 573.34375, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9826396584510804, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.781400297768414e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.84833068847657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5261199951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22300, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.30000715553760526, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.011911691073328257, + "grad_norm": 0.07546477019786835, + "kept_token_count": 572.828125, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9826846480369568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.763832570868544e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7262176513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6806945800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2994038939476013, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.01164669282734394, + "grad_norm": 0.11696706712245941, + "kept_token_count": 573.2078125, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9822582006454468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.930413073860108e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.71463775634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9733154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22320, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3005262613296509, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013657920248806476, + "grad_norm": 0.07279554009437561, + "kept_token_count": 572.5421875, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9815384030342102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.211579031718429e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.38806533813477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.98212280273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22330, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.2993573695421219, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.01236478821374476, + "grad_norm": 0.07395155727863312, + "kept_token_count": 573.0234375, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.983736127614975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.353098069666885e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99715423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8705169677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22340, + "valid_token_count": 817.8546875 + }, + { + "actual_token_mask_ratio": 0.3008822798728943, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.0134440743830055, + "grad_norm": 0.12011201679706573, + "kept_token_count": 572.18125, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9830801844596863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.609323172597214e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.68027038574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8555480957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3003275394439697, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.010629532113671302, + "grad_norm": 0.08240371197462082, + "kept_token_count": 572.4203125, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9831335067749023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.58848905004561e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.64295501708985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.29840087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22360, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.30063243508338927, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.012958945240825415, + "grad_norm": 0.10002033412456512, + "kept_token_count": 572.1546875, + "learning_rate": 5.001191225212096e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9827256500720978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.747817897121423e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.77341766357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1002716064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22370, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.29963299036026003, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.0116421390324831, + "grad_norm": 0.08798574656248093, + "kept_token_count": 572.884375, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9821244597434997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.982654122111853e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77423934936523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0639678955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22380, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2994842916727066, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.013191187009215354, + "grad_norm": 0.09523969888687134, + "kept_token_count": 573.515625, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9832313895225525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.55025738524273e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.95484008789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.89019165039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22390, + "valid_token_count": 818.7109375 + }, + { + "actual_token_mask_ratio": 0.2998910129070282, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.012349927239120007, + "grad_norm": 0.09853823482990265, + "kept_token_count": 573.090625, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9823924660682678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.877961604914162e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.33165740966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3047790527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22400, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.30087172985076904, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.01311623752117157, + "grad_norm": 0.09141291677951813, + "kept_token_count": 571.8015625, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9829498112201691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.660250110144262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.48442993164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6964477539062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22410, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.3001263469457626, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.013188235368579626, + "grad_norm": 0.08045919984579086, + "kept_token_count": 572.9765625, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9832346260547637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.54899366054451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.13503341674804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.11118469238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22420, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.2993952423334122, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.011405973276123405, + "grad_norm": 0.09800653904676437, + "kept_token_count": 573.4375, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9832905352115631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.527150690089911e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9338279724121, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.69366149902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22430, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.30057021975517273, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.012471425719559193, + "grad_norm": 0.1244785338640213, + "kept_token_count": 572.26875, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9833828032016754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.491117455880158e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.05083618164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7345489501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22440, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.3001172125339508, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.013358835782855748, + "grad_norm": 0.06985783576965332, + "kept_token_count": 572.6265625, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9831535756587982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.580654116987717e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.59687042236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4089691162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22450, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.3003394424915314, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.01328166234306991, + "grad_norm": 0.06786968559026718, + "kept_token_count": 572.528125, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9840092420578003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.246408738661558e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.14413604736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.06531677246096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29934456944465637, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010217964788898826, + "grad_norm": 0.09199825674295425, + "kept_token_count": 573.2203125, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9835829555988311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.412922630261164e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.30874328613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4272491455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22470, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.29962305426597596, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.010451958840712905, + "grad_norm": 0.08662259578704834, + "kept_token_count": 573.09375, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9827014625072479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.757259725418408e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.83200378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4226409912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22480, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.3001838237047195, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.010147646069526672, + "grad_norm": 0.09507346153259277, + "kept_token_count": 572.675, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9828083217144012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.715519557474182e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.53838806152343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.35411071777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22490, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.29979206919670104, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.011205740831792354, + "grad_norm": 0.09589073807001114, + "kept_token_count": 572.934375, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9835243463516236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.435825125663541e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.37311706542968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22500, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29972602128982545, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.012431853543967008, + "grad_norm": 0.10701221972703934, + "kept_token_count": 572.884375, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9816753983497619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.1580692747375e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61655654907227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7261627197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22510, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.29972782731056213, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013467434141784907, + "grad_norm": 0.11648411303758621, + "kept_token_count": 573.015625, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9823078095912934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.911029886396136e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5794906616211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9651611328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22520, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.3001592606306076, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.009826760878786445, + "grad_norm": 0.08234354853630066, + "kept_token_count": 572.6578125, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9825990200042725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.797277092118748e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.80001068115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5715301513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2993500232696533, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.012737245997413993, + "grad_norm": 0.12206252664327621, + "kept_token_count": 573.0078125, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9832922995090485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.526467477669939e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.52373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3359741210937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22540, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.2994270116090775, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.011941255675628782, + "grad_norm": 0.07393240928649902, + "kept_token_count": 572.9515625, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9824429452419281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.858235719846562e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2602767944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.32785949707034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22550, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.29999240934848787, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.010346916085109115, + "grad_norm": 0.08577273041009903, + "kept_token_count": 572.771875, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9835979223251343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.407080763892736e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.91464614868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0916809082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22560, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3006942838430405, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.011373142991214991, + "grad_norm": 0.0856783464550972, + "kept_token_count": 572.15, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9820178925991059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.024287042440846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.42272720336913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.86114501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.299306657910347, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.01252370048314333, + "grad_norm": 0.10620579123497009, + "kept_token_count": 573.521875, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9819787859916687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.03955942299217e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.39687728881836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3633758544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22580, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.29903863966464994, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.011569488001987338, + "grad_norm": 0.09923765808343887, + "kept_token_count": 573.446875, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9829397797584534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.664166503469459e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.6235122680664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7354370117188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22590, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29975765645504, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.010736543592065573, + "grad_norm": 0.09376547485589981, + "kept_token_count": 572.934375, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.982925820350647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66962096147472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.18787841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6214111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22600, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.30040846168994906, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.013499552477151155, + "grad_norm": 0.08850560337305069, + "kept_token_count": 572.646875, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.981718385219574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.141280148061923e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42565155029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1846923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22610, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.29973055720329284, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.012952218763530254, + "grad_norm": 0.09744879603385925, + "kept_token_count": 572.8953125, + "learning_rate": 4.696891434591464e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9825824201107025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.80375800584443e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.25545959472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.72222900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22620, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2994024157524109, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.012650296837091447, + "grad_norm": 0.08581287413835526, + "kept_token_count": 573.4421875, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9815369486808777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.212152777356096e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.91869583129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8325958251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22630, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.3003761410713196, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.010868509206920863, + "grad_norm": 0.09721219539642334, + "kept_token_count": 572.3546875, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9821044743061066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.990455294726416e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.08697509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8221374511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22640, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.3003134995698929, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.012679042946547269, + "grad_norm": 0.07513723522424698, + "kept_token_count": 572.2234375, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9829086422920227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.67633161356207e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.66745147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.89356384277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22650, + "valid_token_count": 817.83125 + }, + { + "actual_token_mask_ratio": 0.2994836330413818, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.012755950214341283, + "grad_norm": 0.11212915182113647, + "kept_token_count": 573.2640625, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9823413252830505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.89794440404512e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.31894226074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9053466796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22660, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.30052290856838226, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.012922353856265546, + "grad_norm": 0.08025876432657242, + "kept_token_count": 572.409375, + "learning_rate": 4.6369630696552e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9827330231666564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.744932543369941e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9905014038086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9221527099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22670, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2994777262210846, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.011638179561123252, + "grad_norm": 0.10494720190763474, + "kept_token_count": 573.2453125, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9826836585998535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.764213649148587e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.78624114990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.88629150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22680, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30014916360378263, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.01379777300171554, + "grad_norm": 0.06391047686338425, + "kept_token_count": 572.725, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9826574206352234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.774460634915159e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.480615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.01564636230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22690, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2994586855173111, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.0121509263291955, + "grad_norm": 0.09930147230625153, + "kept_token_count": 573.0921875, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9826837182044983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.764189165551216e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.91238327026367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.34255065917966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22700, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2993871092796326, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.013164084125310182, + "grad_norm": 0.10201133042573929, + "kept_token_count": 573.1515625, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9829997837543487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.640725005127023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.37700500488282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.53175354003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22710, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.30124821364879606, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.013073962787166238, + "grad_norm": 0.08090970665216446, + "kept_token_count": 571.9, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9824051141738892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.873019083286636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.58767395019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3333221435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22720, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.3009343326091766, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.010491958074271678, + "grad_norm": 0.1894376575946808, + "kept_token_count": 572.025, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9819047331809998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.068475897540339e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.31256561279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3308898925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22730, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3000022262334824, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.01330356728285551, + "grad_norm": 0.07979585975408554, + "kept_token_count": 572.5625, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9836246192455291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.396644057531376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91619415283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8348419189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22740, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.29974778890609743, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.012317756982520222, + "grad_norm": 0.09292194992303848, + "kept_token_count": 573.0765625, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9830248057842255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.630952484556473e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.34472732543945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6778991699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22750, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3001991480588913, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012433740124106408, + "grad_norm": 0.08372627943754196, + "kept_token_count": 572.528125, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9827300786972046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.746080034645275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.25729064941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.90233459472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22760, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2999063104391098, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.01203584843315184, + "grad_norm": 0.08581887185573578, + "kept_token_count": 572.8046875, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9831415593624115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.585336959687992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.46531448364257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4308166503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2998749345541, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.01249501588754356, + "grad_norm": 0.10830877721309662, + "kept_token_count": 572.909375, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9815869569778443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.192605989985168e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.33507766723633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1119659423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3002784699201584, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.011954049672931432, + "grad_norm": 0.08861111104488373, + "kept_token_count": 572.3890625, + "learning_rate": 4.494422827645115e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9837566554546356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.345076690195128e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.95902099609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6456237792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22790, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.3003488749265671, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.009763373620808124, + "grad_norm": 0.08042999356985092, + "kept_token_count": 572.68125, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9839571595191956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.266752170631662e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.59823226928711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.692041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22800, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.30011042952537537, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.011408108891919256, + "grad_norm": 0.09046152234077454, + "kept_token_count": 572.63125, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9833226919174194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.514594933832995e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.0924446105957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.97858276367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22810, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.300599130988121, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.010261856112629175, + "grad_norm": 0.06883186846971512, + "kept_token_count": 572.3265625, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9832651317119598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.537073422805407e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90882186889648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8841827392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22820, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.3001554310321808, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.00943520264700055, + "grad_norm": 0.059786830097436905, + "kept_token_count": 572.9609375, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.9835913300514221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.409650814021006e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11193008422852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2868988037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22830, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.29981665015220643, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.012210004357621073, + "grad_norm": 0.1352100968360901, + "kept_token_count": 572.7390625, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9815684258937836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.199850224424154e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.64145965576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3636749267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22840, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.30049667954444886, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.011243696603924037, + "grad_norm": 0.121182382106781, + "kept_token_count": 572.44375, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9827649891376495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.732446345267818e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.842431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.13470458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30047603249549865, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.011183599475771189, + "grad_norm": 0.05727192759513855, + "kept_token_count": 572.5296875, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9835671842098236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.419088604161515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.31911544799804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.30153503417966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22860, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.3002712517976761, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.013643105514347553, + "grad_norm": 0.1051173210144043, + "kept_token_count": 572.284375, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.982072240114212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.003050195635297e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23586807250976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.09332275390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22870, + "valid_token_count": 817.8671875 + }, + { + "actual_token_mask_ratio": 0.3006288677453995, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.012925613019615412, + "grad_norm": 0.06447111815214157, + "kept_token_count": 572.221875, + "learning_rate": 4.388722594512838e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9829568922519684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.657482299488038e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.97453689575195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.17833251953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22880, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2998359352350235, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.012460775254294277, + "grad_norm": 0.12376444041728973, + "kept_token_count": 572.7203125, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.983410918712616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.480123593064491e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.43982696533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8522064208984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22890, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.2999945908784866, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.01153210219927132, + "grad_norm": 0.051861830055713654, + "kept_token_count": 572.771875, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9830800950527191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.609355041291565e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.69989852905273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4639038085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22900, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30101191699504853, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.011589199071750045, + "grad_norm": 0.0829344391822815, + "kept_token_count": 571.925, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9838836550712585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.295469720498658e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.74853439331055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.08867797851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.29980499148368833, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.012232155166566372, + "grad_norm": 0.07854846119880676, + "kept_token_count": 572.7875, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9833488702774048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.504365010187029e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.68794250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5543243408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22920, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.29862392246723174, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.012835207395255566, + "grad_norm": 0.06754360347986221, + "kept_token_count": 573.95, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9828026354312897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.717737451253925e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.200146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.64801330566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22930, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.300398924946785, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.011885380977764726, + "grad_norm": 0.11707746237516403, + "kept_token_count": 572.434375, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.983644700050354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.38881218037568e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.11624450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5424377441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22940, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29953921735286715, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.010279033472761512, + "grad_norm": 0.0859692245721817, + "kept_token_count": 573.165625, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9826549887657166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.775413858122193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.38860855102538, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.36827697753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30058363676071165, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.010207203449681401, + "grad_norm": 0.10423712432384491, + "kept_token_count": 572.240625, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9833930552005767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.487110658781603e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.75791473388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3551300048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22960, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30055853724479675, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.011861253948882221, + "grad_norm": 0.14980359375476837, + "kept_token_count": 572.3953125, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9824586689472199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.852095975773409e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.91164855957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.36705627441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22970, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.3007098317146301, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.012219729553908109, + "grad_norm": 0.0835966244339943, + "kept_token_count": 572.725, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9818485319614411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.090442049957346e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.63811569213867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5176544189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22980, + "valid_token_count": 819.003125 + }, + { + "actual_token_mask_ratio": 0.30028756856918337, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.013135602418333291, + "grad_norm": 0.1460767686367035, + "kept_token_count": 572.6484375, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9822417199611664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.936848512850702e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86826629638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1847442626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22990, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.30093105137348175, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.011195418937131763, + "grad_norm": 0.08338820934295654, + "kept_token_count": 571.978125, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9827513217926025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.737785624864046e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17320251464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9434295654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23000, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.30011661648750304, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.011511127417907118, + "grad_norm": 0.11930184066295624, + "kept_token_count": 572.603125, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9826889276504517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.762152988812887e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.63931198120117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0714935302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23010, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.3008653223514557, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.011327293328940868, + "grad_norm": 0.0859876424074173, + "kept_token_count": 572.034375, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9830862641334533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.606947099498938e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.09496154785157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.88486938476564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23020, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30009943544864653, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.012597585236653686, + "grad_norm": 0.089137502014637, + "kept_token_count": 572.6125, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9825155854225158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.829870253568515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.03476791381836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4252593994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23030, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29937483072280885, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.01258705947548151, + "grad_norm": 0.09367755055427551, + "kept_token_count": 573.3171875, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9830799043178559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.609436713915784e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.93042297363282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9613922119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23040, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.3004458129405975, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.014226073585450649, + "grad_norm": 0.12046388536691666, + "kept_token_count": 572.421875, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9828328132629395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.705950945615769e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.08823471069336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1314758300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23050, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.29921174645423887, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.011460669152438641, + "grad_norm": 0.09762756526470184, + "kept_token_count": 573.5609375, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9822215735912323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.944717388250865e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.46695556640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.43390502929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23060, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.2998677909374237, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.011468372214585543, + "grad_norm": 0.08123025298118591, + "kept_token_count": 573.1140625, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9827096939086915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.754041605745442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49924240112304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.92938842773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23070, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.3003178536891937, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.012012102454900742, + "grad_norm": 0.0870092511177063, + "kept_token_count": 572.634375, + "learning_rate": 4.157587347016891e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9817082405090332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.145238123484887e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.11117401123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6284210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23080, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.29961296916007996, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.011659008311107754, + "grad_norm": 0.09467104822397232, + "kept_token_count": 573.0921875, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9825408399105072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.820002381573431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.50467147827149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3727203369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23090, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.2999071925878525, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.012471091141924262, + "grad_norm": 0.09052456170320511, + "kept_token_count": 572.965625, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.983119934797287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.593791986233554e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34586181640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20436706542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23100, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2992521017789841, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.012874805182218552, + "grad_norm": 0.09131595492362976, + "kept_token_count": 573.5421875, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9827387392520904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.742701079929248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.90479888916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8838592529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23110, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.2993071347475052, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.011227650055661797, + "grad_norm": 0.07535360753536224, + "kept_token_count": 573.365625, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9829684615135192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.65295843646163e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.76874313354492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0756134033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23120, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.29908016622066497, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.010822790162637829, + "grad_norm": 0.10242605209350586, + "kept_token_count": 573.64375, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9830953359603882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.603399706364143e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.21524963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.687451171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23130, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2990560710430145, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.010462506301701069, + "grad_norm": 0.07436266541481018, + "kept_token_count": 573.7265625, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9820400536060333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.015626251813956e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07649383544921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9365631103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23140, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.3008305788040161, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.013477139919996262, + "grad_norm": 0.1141865998506546, + "kept_token_count": 572.2453125, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.982833856344223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.70554261887446e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.37694473266602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6393524169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3000539392232895, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.013690621592104435, + "grad_norm": 0.1340700089931488, + "kept_token_count": 572.771875, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9824343502521515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.861600631964393e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7820556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.19744873046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23160, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.29882229268550875, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.010520808259025216, + "grad_norm": 0.07911019772291183, + "kept_token_count": 573.4625, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9832165598869324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.556050502695143e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.95589447021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1069061279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23170, + "valid_token_count": 817.85625 + }, + { + "actual_token_mask_ratio": 0.2994659125804901, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.01162855066359043, + "grad_norm": 0.06930595636367798, + "kept_token_count": 573.3421875, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9830540180206299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.619536361540668e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9387336730957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4719757080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23180, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2998101443052292, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.012829373683780431, + "grad_norm": 0.10377845913171768, + "kept_token_count": 573.1953125, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9827752590179444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.72842961648712e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.89833221435546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8844268798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23190, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.30137058794498445, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.010626556631177664, + "grad_norm": 0.11873037368059158, + "kept_token_count": 571.8078125, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9829196870326996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.67201580654364e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.03954010009765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.878076171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.29911619126796724, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.010268318653106689, + "grad_norm": 0.0923914760351181, + "kept_token_count": 573.74375, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9833133220672607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.518256086565088e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.94346313476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6916900634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23210, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.300943523645401, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.012380167189985514, + "grad_norm": 0.09493941068649292, + "kept_token_count": 572.0171875, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.982511454820633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831486753071659e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.42041320800782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.28602905273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23220, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.30010275840759276, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.011026954278349876, + "grad_norm": 0.09091351181268692, + "kept_token_count": 572.56875, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.982781183719635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.726117280777543e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.290087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1183624267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23230, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.29995841681957247, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.012254845630377532, + "grad_norm": 0.09361062198877335, + "kept_token_count": 572.9671875, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9830779731273651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.610185337194708e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48219146728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.05991516113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23240, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.299342480301857, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.011514703929424285, + "grad_norm": 0.0647728219628334, + "kept_token_count": 573.2640625, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9818846046924591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.076343208609615e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.2497688293457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1729217529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23250, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3000176638364792, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.014093586523085833, + "grad_norm": 0.06793271750211716, + "kept_token_count": 572.6875, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9819897532463073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.035268572508357e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22351989746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.39110107421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23260, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.29976696372032163, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.010236158268526196, + "grad_norm": 0.0709223747253418, + "kept_token_count": 572.9625, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9823817014694214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.882165944261942e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.29182434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.65517272949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23270, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30068847835063933, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.01249822243116796, + "grad_norm": 0.07898537069559097, + "kept_token_count": 572.1484375, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9830828666687011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.608275762118865e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.18496017456054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.40863037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.3002413183450699, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.010740291234105826, + "grad_norm": 0.08237063139677048, + "kept_token_count": 572.76875, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9825244843959808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.826393146184274e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.84642181396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7450378417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23290, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.3007210224866867, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.011345696821808814, + "grad_norm": 0.08990832418203354, + "kept_token_count": 572.2265625, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9829521298408508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.659341124759521e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.82145843505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6899169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23300, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.30055749118328096, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.01181566077284515, + "grad_norm": 0.0849800631403923, + "kept_token_count": 572.403125, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.982911616563797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.675172116956673e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.2560432434082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.85191040039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23310, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3007942736148834, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.00964811909943819, + "grad_norm": 0.09396828711032867, + "kept_token_count": 572.2609375, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9834222078323365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.475720329035539e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44576873779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8954711914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23320, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.3004162013530731, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.010850069299340247, + "grad_norm": 0.06642209738492966, + "kept_token_count": 572.434375, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9826949775218964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.759793468518183e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.70867919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.21286926269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23330, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2991498738527298, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.013446673937141895, + "grad_norm": 0.1030607521533966, + "kept_token_count": 573.4078125, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9834877848625183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.450099826906807e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.64625396728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7236297607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23340, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3002489537000656, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.011039721500128507, + "grad_norm": 0.10256058722734451, + "kept_token_count": 572.546875, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9826194941997528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.789281142118852e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.63004684448242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.70958251953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2991775631904602, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.011655878275632858, + "grad_norm": 0.07479409873485565, + "kept_token_count": 573.2640625, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9827018618583679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.757106239092536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.23342895507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3452667236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23360, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.3003643035888672, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.013140894658863544, + "grad_norm": 0.11563481390476227, + "kept_token_count": 572.4625, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9819661557674408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.044494595902507e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1661750793457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1260498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23370, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29983487725257874, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.011951506230980158, + "grad_norm": 0.07255648076534271, + "kept_token_count": 572.9921875, + "learning_rate": 3.820799398012889e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9835271298885345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.434744354919531e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.36931610107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4879943847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23380, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30030760169029236, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.011497814627364277, + "grad_norm": 0.09634703397750854, + "kept_token_count": 572.3015625, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9829474747180938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.661161132797133e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.90837249755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8012969970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23390, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.29956677854061126, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.01103688832372427, + "grad_norm": 0.11732892692089081, + "kept_token_count": 573.1609375, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9834430634975433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.467580424214247e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22125701904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1736724853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23400, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2990116775035858, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.011460894579067826, + "grad_norm": 0.060374915599823, + "kept_token_count": 573.575, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9831497609615326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.582141795661301e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75835876464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8801696777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.2989650428295135, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.0128300363663584, + "grad_norm": 0.07279328256845474, + "kept_token_count": 573.821875, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9833407521247863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.507537291327026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.60079498291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9418670654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23420, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.299328950047493, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.012222363473847508, + "grad_norm": 0.06824450194835663, + "kept_token_count": 573.1890625, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9831178605556488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.594598235096783e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49438858032227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1056823730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23430, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.2995860606431961, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.011376608442515136, + "grad_norm": 0.09880440682172775, + "kept_token_count": 572.896875, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.983632093667984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.393732983269728e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.76930618286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.423779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23440, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30036832094192506, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.010690717957913876, + "grad_norm": 0.11293644458055496, + "kept_token_count": 572.596875, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9825751960277558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.806577221141197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1784896850586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8315185546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23450, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.2991848886013031, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.013842819817364216, + "grad_norm": 0.08714516460895538, + "kept_token_count": 573.346875, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.983021742105484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.63215647364268e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.20805969238282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1555572509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23460, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29997862577438356, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.013429825566709042, + "grad_norm": 0.06692299991846085, + "kept_token_count": 572.6734375, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9835248947143554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.435609138861764e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.27127456665039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.452978515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23470, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.30094325840473174, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.012848148308694362, + "grad_norm": 0.08692307025194168, + "kept_token_count": 572.034375, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9819584906101226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.04748323187232e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.90038452148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9300994873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.30076721012592317, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.010847873333841563, + "grad_norm": 0.0692594051361084, + "kept_token_count": 572.153125, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9826179265975952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.789889994252007e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23928146362304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.17713317871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23490, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2996494144201279, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.012545181205496192, + "grad_norm": 0.07711299508810043, + "kept_token_count": 572.9671875, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9833936274051667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.48688288492849e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.00902938842773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.99537963867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23500, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.2993947505950928, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.011154972529038788, + "grad_norm": 0.08143428713083267, + "kept_token_count": 573.2375, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.983708244562149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.363986140058841e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.56050720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4557342529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23510, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.3005756586790085, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.0126311588101089, + "grad_norm": 0.10817699134349823, + "kept_token_count": 572.4484375, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.983050549030304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.620899002882653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76565093994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4017333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23520, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3005619585514069, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.012724124686792493, + "grad_norm": 0.08608007431030273, + "kept_token_count": 572.4984375, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9839698195457458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.261803027882707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47384490966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6967041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23530, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.3004184186458588, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.013047125842422246, + "grad_norm": 0.10790926963090897, + "kept_token_count": 572.7890625, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9830968976020813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.602795401704498e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.68850708007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8594451904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23540, + "valid_token_count": 818.759375 + }, + { + "actual_token_mask_ratio": 0.29953503608703613, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.012328245956450701, + "grad_norm": 0.08577272295951843, + "kept_token_count": 573.2171875, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9830125868320465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.635728168475907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.4813735961914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8862823486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23550, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2995228976011276, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.012172689381986856, + "grad_norm": 0.12429153174161911, + "kept_token_count": 572.9640625, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9824693024158477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.847948170616292e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.41926040649415, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9827484130859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23560, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.30020272731781006, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.012785499170422553, + "grad_norm": 0.0966704785823822, + "kept_token_count": 572.8125, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9833886504173279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.488829203590284e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.40422973632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9939819335938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23570, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2997092425823212, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.011249324772506952, + "grad_norm": 0.10424867272377014, + "kept_token_count": 573.1296875, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9825435519218445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.818941692472435e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.71902847290039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9003021240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23580, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.2990772634744644, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.011507781967520715, + "grad_norm": 0.07161448150873184, + "kept_token_count": 573.6171875, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9823426246643067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.897427374497056e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10059967041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.20078735351564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23590, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.30004534125328064, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.011652746330946684, + "grad_norm": 0.07242289185523987, + "kept_token_count": 572.78125, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.98345587849617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.462568417191506e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.87831344604493, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9851928710938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23600, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.2989743500947952, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.011482310621067882, + "grad_norm": 0.10385457426309586, + "kept_token_count": 573.4765625, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9833209931850433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.515250061056577e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.13580017089843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9220336914062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23610, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.30096291899681094, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.013410544581711292, + "grad_norm": 0.10788875818252563, + "kept_token_count": 571.9390625, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9833351850509644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.50970840069931e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.81180191040039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4943328857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23620, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2996260583400726, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.01108063543215394, + "grad_norm": 0.1302071362733841, + "kept_token_count": 573.2859375, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9838065266609192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.32559309451608e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.60203323364257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7429992675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23630, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.29965334534645083, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.010619634576141835, + "grad_norm": 0.09871665388345718, + "kept_token_count": 573.06875, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9833985805511475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.484951045422349e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.3724136352539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5610595703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.3002388715744019, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.011678466899320483, + "grad_norm": 0.08889148384332657, + "kept_token_count": 572.5421875, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9832509815692901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.542602277477272e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.17995834350586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.117724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23650, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2995465397834778, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.01404456477612257, + "grad_norm": 0.08848016709089279, + "kept_token_count": 573.1078125, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9841159820556641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.204714009072632e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.44368515014648, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9984558105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23660, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.29984750747680666, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.01122169140726328, + "grad_norm": 0.08440741896629333, + "kept_token_count": 572.996875, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9832582652568818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.539757414429914e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.90945358276367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.51304016113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23670, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2999254584312439, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.010313673689961434, + "grad_norm": 0.0862262099981308, + "kept_token_count": 572.771875, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9821612179279328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.968294983380474e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.2781753540039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.832470703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23680, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2998274266719818, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.011478625005111098, + "grad_norm": 0.08621600270271301, + "kept_token_count": 573.025, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9828408241271973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.702818427584134e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.05973587036132, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.76536560058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23690, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.29995604157447814, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.010760621493682265, + "grad_norm": 0.08417099714279175, + "kept_token_count": 572.6734375, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9829819023609161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.647711525147315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.26377029418946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.31358642578124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23700, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.30099269449710847, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.011005765106528998, + "grad_norm": 0.09359818696975708, + "kept_token_count": 571.803125, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9837478756904602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.348503375193104e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.08010330200196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1660430908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23710, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.30030505955219267, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.012177598802372814, + "grad_norm": 0.10607963055372238, + "kept_token_count": 572.5328125, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9824647784233094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.849719939054922e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.72416687011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0897491455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23720, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29986886978149413, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.011983567290008068, + "grad_norm": 0.08342289179563522, + "kept_token_count": 572.834375, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9832665205001831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.536528999276925e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.78698348999023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4665100097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23730, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.30057367086410525, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.011661671847105027, + "grad_norm": 0.08330047875642776, + "kept_token_count": 572.425, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9836899757385253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.371123708959203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.6949478149414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3344421386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23740, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.30007362365722656, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.013085620198398828, + "grad_norm": 0.09134336560964584, + "kept_token_count": 572.915625, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9827520251274109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.737507465004455e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.55553588867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92719116210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23750, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2999114513397217, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.01296503683552146, + "grad_norm": 0.10290464013814926, + "kept_token_count": 572.853125, + "learning_rate": 3.4117955855922e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9830921113491058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.604662994504906e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.65177383422852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6326568603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2998675972223282, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.011037996504455805, + "grad_norm": 0.08659879863262177, + "kept_token_count": 572.86875, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9819908857345581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.034831323835534e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.23484268188477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9726867675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23770, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.299713534116745, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.01124874665401876, + "grad_norm": 0.08745558559894562, + "kept_token_count": 573.0140625, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9829109311103821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.675438162346836e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.46326141357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.29587707519534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23780, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.3008762001991272, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.011451504146680235, + "grad_norm": 0.07725291699171066, + "kept_token_count": 571.9703125, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9824302673339844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.863195303594694e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.72998886108398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9598663330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23790, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2985044181346893, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.011354849161580204, + "grad_norm": 0.09360544383525848, + "kept_token_count": 573.85625, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.98339142203331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.487748687504791e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.41918411254883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.17832641601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23800, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.3004759341478348, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.012393403425812721, + "grad_norm": 0.12219098955392838, + "kept_token_count": 572.509375, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9827861309051513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.724186314386316e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.69692916870117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.31932373046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23810, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2999488890171051, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.012085232231765986, + "grad_norm": 0.08616971969604492, + "kept_token_count": 572.74375, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9825988054275513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.797361857024953e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.53775253295899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23820, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2987163096666336, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.013733479846268892, + "grad_norm": 0.10539925843477249, + "kept_token_count": 573.8546875, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9829160213470459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.673446041531861e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.81570816040039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9996826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23830, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29927034080028536, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.011830421863123775, + "grad_norm": 0.0886501744389534, + "kept_token_count": 573.5265625, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9818873703479767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.075262947182636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8477569580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1766876220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23840, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.30022100508213045, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.012951169814914464, + "grad_norm": 0.155095174908638, + "kept_token_count": 572.5984375, + "learning_rate": 3.31787790672602e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9824623048305512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.850678328191861e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.93961639404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.55845336914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23850, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29914698004722595, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012151765637099743, + "grad_norm": 0.08893215656280518, + "kept_token_count": 573.390625, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9830315053462982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.628331757383422e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.67649688720704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.69928588867185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23860, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2999108970165253, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.010459378687664866, + "grad_norm": 0.096479631960392, + "kept_token_count": 572.9984375, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9833004653453827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.523268821183592e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.82949371337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5504577636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23870, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2994652301073074, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.010188629431650043, + "grad_norm": 0.08772820234298706, + "kept_token_count": 573.2390625, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9827484250068664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.738913798471912e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.89411926269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.49220886230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23880, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3006118506193161, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.010868321126326918, + "grad_norm": 0.10734055191278458, + "kept_token_count": 572.4984375, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9823652386665345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.88859639922157e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.1299331665039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1035675048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23890, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.30086180567741394, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.01183352041989565, + "grad_norm": 0.07439488172531128, + "kept_token_count": 572.0875, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.983579283952713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.414363197109197e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.56015396118164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.76048278808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23900, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.30060683488845824, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.010856552887707949, + "grad_norm": 0.07325015962123871, + "kept_token_count": 572.4921875, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9845424294471741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.038130377419293e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.85035934448243, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6251525878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23910, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.29963305592536926, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.009617578890174627, + "grad_norm": 0.11433959007263184, + "kept_token_count": 573.340625, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9829425096511841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.663107451458928e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.64480514526367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6290740966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23920, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.29984181225299833, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.011459605069831014, + "grad_norm": 0.10132942348718643, + "kept_token_count": 572.978125, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9825460612773895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.817963148932903e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.75054168701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5896026611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23930, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.2992196947336197, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.011446426622569561, + "grad_norm": 0.07641655206680298, + "kept_token_count": 573.5453125, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9824347198009491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.861460860818624e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.86892242431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.22272033691405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23940, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.29990815818309785, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.012615693546831609, + "grad_norm": 0.12063135951757431, + "kept_token_count": 573.090625, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9826799869537354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.765641664969735e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.5759407043457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6790832519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23950, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.2994218349456787, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.013823444768786431, + "grad_norm": 0.09975706785917282, + "kept_token_count": 573.503125, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9827534317970276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.736957984685433e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08752365112305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4680450439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23960, + "valid_token_count": 818.61875 + }, + { + "actual_token_mask_ratio": 0.30025621950626374, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.012943568453192712, + "grad_norm": 0.4669688642024994, + "kept_token_count": 572.5109375, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9825108706951141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831709033576772e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.44574508666992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6527160644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3001319825649261, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.011252030963078141, + "grad_norm": 0.09969226270914078, + "kept_token_count": 572.6546875, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9837149500846862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.361365813063457e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.47819366455079, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.741748046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23980, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.3006003886461258, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.01163086467422545, + "grad_norm": 0.09238046407699585, + "kept_token_count": 572.39375, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9824432671070099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.858114356873557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.30654067993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2052001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29914675951004027, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.010614759661257267, + "grad_norm": 0.09239176660776138, + "kept_token_count": 573.4875, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.982112032175064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.987504930293653e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.05026397705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0635589599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24000, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2998869061470032, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.013054110342636704, + "grad_norm": 0.06519744545221329, + "kept_token_count": 572.6796875, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9829291224479675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.668329697276932e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.76195678710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1453552246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24010, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.29989157021045687, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.011776102427393197, + "grad_norm": 0.08337096869945526, + "kept_token_count": 572.965625, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9827271223068237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.74723123665899e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.51360855102538, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1474548339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24020, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3005417823791504, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.010934811271727084, + "grad_norm": 0.17969098687171936, + "kept_token_count": 572.428125, + "learning_rate": 3.133501007254138e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9826399862766266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.781271149520762e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.81598663330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3828887939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24030, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.30011809170246123, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.0094494195189327, + "grad_norm": 0.0939243733882904, + "kept_token_count": 572.6125, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.98327516913414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.533152372867335e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.29351043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2704315185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24040, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2995555490255356, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.011806735023856162, + "grad_norm": 0.1403750777244568, + "kept_token_count": 573.215625, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9819784760475159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.039675510895904e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9838134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.447119140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24050, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.3007413327693939, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.0110732801258564, + "grad_norm": 0.11193259060382843, + "kept_token_count": 572.003125, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9825515210628509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.815832603024319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.86502227783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1601928710937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24060, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.29997736513614653, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.01190461958758533, + "grad_norm": 0.09620335698127747, + "kept_token_count": 572.7984375, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9824454247951507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.857275729998946e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.37470932006836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0661651611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24070, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2996220737695694, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.011423078319057822, + "grad_norm": 0.11762886494398117, + "kept_token_count": 573.2328125, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9832115828990936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.557996166520753e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55373687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0355682373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.30072916150093076, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.01119452021084726, + "grad_norm": 0.10099410265684128, + "kept_token_count": 571.990625, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9832137167453766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.557162196259014e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.56642684936523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.50731201171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24090, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.30025743544101713, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.011857766518369317, + "grad_norm": 0.1303892880678177, + "kept_token_count": 572.3390625, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9834366142749786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.470093248935881e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.04969100952148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20604248046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24100, + "valid_token_count": 817.921875 + }, + { + "actual_token_mask_ratio": 0.29981801509857176, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.01003520847298205, + "grad_norm": 0.07261347770690918, + "kept_token_count": 573.021875, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9826493561267853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.777615089958999e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.8081886291504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2715118408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24110, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.29992839097976687, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.010265082539990544, + "grad_norm": 0.10709843039512634, + "kept_token_count": 572.6078125, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9822201609611512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.945259738131427e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.86776275634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7400756835938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24120, + "valid_token_count": 817.93125 + }, + { + "actual_token_mask_ratio": 0.29958716928958895, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.012394906487315893, + "grad_norm": 0.09921391308307648, + "kept_token_count": 573.0515625, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9828819751739502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.686749766231515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77036361694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9370361328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24130, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2999265789985657, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.010284595238044859, + "grad_norm": 0.11727266758680344, + "kept_token_count": 572.8625, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9828510701656341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.698813340335618e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.0096939086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48919372558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24140, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.29977600276470184, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.01052317195571959, + "grad_norm": 0.07583245635032654, + "kept_token_count": 572.83125, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9820919752120971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.995339135755785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.27830963134765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.62192687988284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24150, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.30042050778865814, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.01299503301270306, + "grad_norm": 0.08914685249328613, + "kept_token_count": 572.2921875, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.013, + "mgd_cos_sim": 0.983122181892395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.592916142835747e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09457778930664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8666137695312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24160, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.3001557469367981, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.011889344034716486, + "grad_norm": 0.09960940480232239, + "kept_token_count": 572.784375, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.983237338066101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.547934317495674e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86725616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6654815673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29964260160923006, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.0123903120867908, + "grad_norm": 0.0960075706243515, + "kept_token_count": 573.00625, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.982586407661438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.802207390137483e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.91547622680665, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9008483886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.29970364570617675, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.010109518142417073, + "grad_norm": 0.09848590195178986, + "kept_token_count": 573.0765625, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9825312972068787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.823725307185669e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.07792663574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.60881958007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24190, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29916932284832, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.013009421527385712, + "grad_norm": 0.10389602929353714, + "kept_token_count": 573.496875, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9818742334842682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.080396171659231e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25526962280273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5336181640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24200, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.3016327202320099, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.012233956856653095, + "grad_norm": 0.12002728134393692, + "kept_token_count": 571.6265625, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9822943091392518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.91630513756536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.9548355102539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1622344970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24210, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2996073216199875, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.011867985734716058, + "grad_norm": 0.15545488893985748, + "kept_token_count": 573.1, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.983009260892868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.637027727265376e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.51117782592773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.75547180175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24220, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30109438896179197, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.009983516065403819, + "grad_norm": 0.06900646537542343, + "kept_token_count": 571.7125, + "learning_rate": 2.934119885218902e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9832056283950805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.560317669936921e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.35901641845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4368408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24230, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.2998563200235367, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.01027244608849287, + "grad_norm": 0.10961300134658813, + "kept_token_count": 572.984375, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9828846454620361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.685705520794726e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7142723083496, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.36412048339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24240, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30002066791057586, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.011337406560778618, + "grad_norm": 0.08176272362470627, + "kept_token_count": 572.8171875, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9826255738735199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.786910053051543e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.18119888305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.13280029296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24250, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29946320950984956, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.010672916378825902, + "grad_norm": 0.07615003734827042, + "kept_token_count": 573.196875, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9829986155033111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.641189174843021e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.6959831237793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8721557617188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24260, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3001491129398346, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.011442843871191145, + "grad_norm": 0.0697653666138649, + "kept_token_count": 572.85, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9830458045005799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.622751971008256e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48692245483399, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.97312316894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24270, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.3004214435815811, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.010705969203263522, + "grad_norm": 0.09368158876895905, + "kept_token_count": 572.6234375, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9837288796901703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.355927944241557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3246940612793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.85858459472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24280, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29976347982883456, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.01336420988664031, + "grad_norm": 0.09920013695955276, + "kept_token_count": 573.0171875, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9820502936840058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.011630077613518e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.80064544677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2667663574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24290, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.3000294387340546, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.01202021325007081, + "grad_norm": 0.110402412712574, + "kept_token_count": 572.6625, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9836702287197113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.378841098921839e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.80082473754882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.57349548339846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24300, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.29926520586013794, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.01302938936278224, + "grad_norm": 0.08894722908735275, + "kept_token_count": 573.3921875, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.983075487613678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.611154021811672e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.57555694580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5293243408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24310, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29932336807250975, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.012285055872052909, + "grad_norm": 0.0920867994427681, + "kept_token_count": 573.19375, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9832866966724396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.528657904709689e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.49522705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.89747314453126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24320, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.30056426525115965, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.010494129965081811, + "grad_norm": 0.0842885971069336, + "kept_token_count": 572.0484375, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9833264768123626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.513108564831782e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.78456802368164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.557421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24330, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.3008504569530487, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.011546243960037828, + "grad_norm": 0.08162326365709305, + "kept_token_count": 572.0734375, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9840131640434265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.244876349228434e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.9136947631836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3070831298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24340, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2997776210308075, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.012181106442585587, + "grad_norm": 0.0869831070303917, + "kept_token_count": 572.98125, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9826963782310486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.759250027243979e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.4488151550293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4679412841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24350, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.29887881875038147, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.011930596362799407, + "grad_norm": 0.09114186465740204, + "kept_token_count": 573.6890625, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9833935260772705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.486918828159105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.1762306213379, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1909240722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3001283198595047, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.011088930815458298, + "grad_norm": 0.08591487258672714, + "kept_token_count": 572.5765625, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9836362838745117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.39209207292879e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.24192276000977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.72803649902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24370, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.2995948553085327, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.010415996750816702, + "grad_norm": 0.07579304277896881, + "kept_token_count": 572.9796875, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9832558870315552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.540687245433219e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9439910888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5050506591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24380, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.3001809924840927, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.011124044749885797, + "grad_norm": 0.12421959638595581, + "kept_token_count": 572.7546875, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9829147577285766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.673939424217679e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.53125610351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.57282104492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24390, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2999907523393631, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.011894859466701746, + "grad_norm": 0.15449866652488708, + "kept_token_count": 572.74375, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9827565252780914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.73574955726508e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5051055908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1854553222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24400, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3003180265426636, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.013579696416854858, + "grad_norm": 0.13284216821193695, + "kept_token_count": 572.4515625, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9827879011631012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.723492879245896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.47599945068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8908386230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24410, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.29963610172271726, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.01089059617370367, + "grad_norm": 0.1007276400923729, + "kept_token_count": 572.8828125, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9831957817077637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.564162722497713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.6255714416504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3723571777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24420, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.29959518313407896, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.013137682434171438, + "grad_norm": 0.1355118602514267, + "kept_token_count": 573.284375, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9818529844284057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.088693091645837e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.80593566894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0879272460937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24430, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.2999863773584366, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.012133279349654913, + "grad_norm": 0.09902350604534149, + "kept_token_count": 572.621875, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9829953789710999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.64244857034646e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.40005569458008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2550750732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24440, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.299788698554039, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.015558748552575708, + "grad_norm": 0.07249822467565536, + "kept_token_count": 573.0515625, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9827217161655426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.749348322045989e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.21306991577148, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.939306640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24450, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.30035478472709654, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.008512877766042948, + "grad_norm": 0.08540348708629608, + "kept_token_count": 572.2234375, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9835902988910675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.41005204670364e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.51110534667968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.37732849121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24460, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.3011000692844391, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.009400812024250627, + "grad_norm": 0.0789463147521019, + "kept_token_count": 571.815625, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0094, + "mgd_cos_sim": 0.9834505617618561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.464646066888236e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7900276184082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.16151428222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24470, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29991120398044585, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.012598653510212898, + "grad_norm": 0.09134698659181595, + "kept_token_count": 572.925, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9829347789287567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.666120534646325e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.70584869384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2801086425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24480, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.29969289898872375, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.009735688846558333, + "grad_norm": 0.06706518679857254, + "kept_token_count": 572.9640625, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9839104831218719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.284989285632037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61118698120117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6300567626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29854128062725066, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.012616979191079736, + "grad_norm": 0.06585181504487991, + "kept_token_count": 573.94375, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9826116681098938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.792334315832704e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.70756454467774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7579711914062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24500, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.3001556843519211, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.01182896145619452, + "grad_norm": 0.0954412966966629, + "kept_token_count": 572.7390625, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9833485901355743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.504475641122553e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.15388946533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0962646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24510, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.3001305401325226, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.010646155430004, + "grad_norm": 0.08344296365976334, + "kept_token_count": 572.86875, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9825922310352325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.799930743000005e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.67868576049804, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6084411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24520, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.3004016399383545, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.010514304134994745, + "grad_norm": 0.0674406886100769, + "kept_token_count": 572.378125, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9844190120697022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.086344437790103e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.47757339477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2510162353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2996954441070557, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.011300452519208193, + "grad_norm": 0.08039099723100662, + "kept_token_count": 573.20625, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9833636343479156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.498598668258637e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.05281372070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4901641845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24540, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2988328903913498, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.011337008140981198, + "grad_norm": 0.07292808592319489, + "kept_token_count": 573.7625, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9828821659088135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.686663073196542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.54588165283204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5321472167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24550, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2994572937488556, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.01154454448260367, + "grad_norm": 0.10596980899572372, + "kept_token_count": 573.109375, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9830565214157104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.618566003453452e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.74185791015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7970825195313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24560, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3014101773500443, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.010692430846393108, + "grad_norm": 0.10435986518859863, + "kept_token_count": 571.7546875, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9836201369762421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.398412006092258e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.60881423950195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9343017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24570, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.29880607426166533, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.011349187744781376, + "grad_norm": 0.08335787802934647, + "kept_token_count": 573.828125, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9824878215789795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.840712740086019e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.06119155883789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.87123107910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24580, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.3003500461578369, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.011088088573887944, + "grad_norm": 0.13586527109146118, + "kept_token_count": 572.465625, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9829554617404938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.658036218141206e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.683740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1191009521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24590, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29965530931949613, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.011657497752457857, + "grad_norm": 0.08433949202299118, + "kept_token_count": 573.1296875, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.983464390039444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.459243304561824e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.64755249023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47117919921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24600, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29945496618747713, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.010221856320276857, + "grad_norm": 0.09834802895784378, + "kept_token_count": 573.275, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9835641980171204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.420247082132846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.08843078613282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.10655517578124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24610, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3012655436992645, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.011365396017208696, + "grad_norm": 0.10411056131124496, + "kept_token_count": 571.6046875, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9829359591007233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.665662513114513e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.65018157958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1240997314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24620, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3002800166606903, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.010536436457186937, + "grad_norm": 0.07777039706707001, + "kept_token_count": 572.50625, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9833513140678406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.503410368168261e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.9392593383789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8370025634766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24630, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.3000830501317978, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.0103511321824044, + "grad_norm": 0.11471579223871231, + "kept_token_count": 572.8, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9835682392120362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.418673037842382e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07023086547852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7950134277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24640, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.3006830394268036, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.00931723597459495, + "grad_norm": 0.07346207648515701, + "kept_token_count": 571.978125, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9833364248275757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.5092290242319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07134475708008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3007537841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24650, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.30000861585140226, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.011969624599441887, + "grad_norm": 0.08619587123394012, + "kept_token_count": 572.78125, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.012, + "mgd_cos_sim": 0.982394403219223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.87719788402319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.47269210815429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7385498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3005991578102112, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.010331414313986898, + "grad_norm": 0.0901007428765297, + "kept_token_count": 572.2390625, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.983635550737381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.392383402271662e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.17483215332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0382110595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24670, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.2996092736721039, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.010293395863845945, + "grad_norm": 0.14706411957740784, + "kept_token_count": 572.9921875, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9824208796024323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.866866315249353e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.20413818359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4471801757812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24680, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.3002243906259537, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.012493712268769741, + "grad_norm": 0.1020929366350174, + "kept_token_count": 572.8171875, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9822227954864502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.944239285076038e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.85268325805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8483917236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24690, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.2995990484952927, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.010939109884202481, + "grad_norm": 0.10455336421728134, + "kept_token_count": 573.0265625, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9835029661655426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.444167993322481e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47787094116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5838928222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24700, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.3000927060842514, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.012968057440593839, + "grad_norm": 0.11397243291139603, + "kept_token_count": 572.5578125, + "learning_rate": 2.479709800807915e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9821393311023712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.97684045007918e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.92239227294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.03493041992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24710, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.2995022863149643, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010325160343199968, + "grad_norm": 0.06732533127069473, + "kept_token_count": 573.06875, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9826785624027252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.766204096493311e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08753204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9939270019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2998934775590897, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.010717593552544712, + "grad_norm": 0.10452491790056229, + "kept_token_count": 572.940625, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9832954704761505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.525229837279767e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.86724395751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.76633605957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24730, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.299731183052063, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.01222508163191378, + "grad_norm": 0.08127619326114655, + "kept_token_count": 573.2625, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9830343723297119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.627216498600319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.55802917480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.62864990234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24740, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.301459863781929, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.01124722664244473, + "grad_norm": 0.13240526616573334, + "kept_token_count": 571.9078125, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9821920812129974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.956238721613772e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.73669204711913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4468994140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24750, + "valid_token_count": 818.7125 + }, + { + "actual_token_mask_ratio": 0.3000719755887985, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.010682778107002378, + "grad_norm": 0.21654698252677917, + "kept_token_count": 572.5625, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9828952074050903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.681578197458294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.79820938110352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9657806396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24760, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.29954020082950594, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.012308182148262858, + "grad_norm": 0.08594828099012375, + "kept_token_count": 573.290625, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9829009473323822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.679333309875802e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34851455688477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1861145019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24770, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29967553317546847, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.010551388282328844, + "grad_norm": 0.0646648183465004, + "kept_token_count": 573.2234375, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9837571442127228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.344890498439781e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.75055160522462, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.12999267578124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24780, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.3006869077682495, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.012422613706439734, + "grad_norm": 0.0886574387550354, + "kept_token_count": 572.31875, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9824067234992981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.87239102262538e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.19698486328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4745727539063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24790, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.29910086989402773, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.011533074826002122, + "grad_norm": 0.09539910405874252, + "kept_token_count": 573.5953125, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9828657567501068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.693084505968728e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.52389602661133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6340789794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24800, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.2998169958591461, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.011716284463182092, + "grad_norm": 0.11679242551326752, + "kept_token_count": 573.0703125, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9831724047660828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.57329845125787e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.32670822143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7405059814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24810, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.30102369785308836, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012354899244382978, + "grad_norm": 0.1253053992986679, + "kept_token_count": 572.0953125, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9833696782588959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.496235509985126e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.25496826171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3765045166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24820, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.30060692727565763, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.011589000467211008, + "grad_norm": 0.11530855298042297, + "kept_token_count": 572.1953125, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9837821364402771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.335126818157732e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.96773223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9888031005859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24830, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2999466061592102, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.010765974689275027, + "grad_norm": 0.0773383304476738, + "kept_token_count": 572.740625, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9829451978206635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.662051237071864e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.88259811401367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3354766845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24840, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2992195636034012, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.01100640194490552, + "grad_norm": 0.15294702351093292, + "kept_token_count": 573.3359375, + "learning_rate": 2.353720475835494e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9828367054462432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.704431689286139e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.54315567016602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.13075256347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24850, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2985359162092209, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.016501353960484265, + "grad_norm": 0.10274291783571243, + "kept_token_count": 574.1875, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9821839332580566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.959414677112363e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.6160903930664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9123168945313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24860, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.2996745526790619, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.011143924156203865, + "grad_norm": 0.1159239411354065, + "kept_token_count": 573.1765625, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9827419102191925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.741465185768902e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.42730255126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1746826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24870, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.2993899255990982, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.01192095447331667, + "grad_norm": 0.13658827543258667, + "kept_token_count": 573.2375, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9834140360355377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.47891174594406e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07636489868165, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4494598388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24880, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3004284143447876, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.011534006567671895, + "grad_norm": 0.09150519967079163, + "kept_token_count": 572.578125, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9823035299777985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.912711905897595e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.04898300170899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1020477294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24890, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.29954011738300323, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.011229126527905465, + "grad_norm": 0.08033335208892822, + "kept_token_count": 573.334375, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9821258604526519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.982099512242712e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.84778213500977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1191009521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24900, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.30051997900009153, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.01276001362130046, + "grad_norm": 0.08501750975847244, + "kept_token_count": 572.465625, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9836292505264282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.39484052953776e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.90021896362305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.29541015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24910, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3001855671405792, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.012579254247248173, + "grad_norm": 0.1151784360408783, + "kept_token_count": 572.7796875, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9828827559947968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.686446358799003e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.901611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.82306518554685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24920, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.3009572744369507, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.012556343339383602, + "grad_norm": 0.09872869402170181, + "kept_token_count": 572.140625, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9827822983264923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.725684688717593e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.85432434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2245300292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24930, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2997830659151077, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.012233050260692834, + "grad_norm": 0.1248898059129715, + "kept_token_count": 572.834375, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9829697072505951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.65248131554108e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.31035842895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2796661376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24940, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2989801377058029, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.010291609866544604, + "grad_norm": 0.08666057884693146, + "kept_token_count": 573.553125, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9832412242889405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.546413424075581e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.76879043579102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1041748046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24950, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2987527519464493, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.012244716053828597, + "grad_norm": 0.09468747675418854, + "kept_token_count": 573.7765625, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9834820210933686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.452355301007628e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.25329895019532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.49285888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24960, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29867281913757326, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.011961124930530786, + "grad_norm": 0.08313074707984924, + "kept_token_count": 573.8265625, + "learning_rate": 2.248125622537373e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9821893572807312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.957297227927484e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.53740615844727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6559295654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2983385056257248, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.010450104344636202, + "grad_norm": 0.0812193974852562, + "kept_token_count": 573.984375, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.98382648229599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.317801562545355e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.58237991333007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.03455810546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24980, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2999820411205292, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.010587806859984995, + "grad_norm": 0.07239057123661041, + "kept_token_count": 572.64375, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9845047533512116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.052845747035462e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.50253067016601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.694140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24990, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.299592062830925, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.014083348587155342, + "grad_norm": 0.08554410934448242, + "kept_token_count": 573.20625, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9834379136562348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.469586915045512e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.73665161132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6655578613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25000, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.30063454806804657, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.010930690681561828, + "grad_norm": 0.10888262093067169, + "kept_token_count": 572.20625, + "learning_rate": 2.213422261096668e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9820756554603577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.001718986430206e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.1579963684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.07831115722655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25010, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2993397444486618, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.010381862288340926, + "grad_norm": 0.11083522439002991, + "kept_token_count": 573.2984375, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9824984312057495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.836569809820503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.06893157958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7346466064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25020, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.29920899868011475, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.0122854876331985, + "grad_norm": 0.06827645748853683, + "kept_token_count": 573.1484375, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9818502247333527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.089780192472971e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.98661956787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.41846923828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25030, + "valid_token_count": 817.859375 + }, + { + "actual_token_mask_ratio": 0.3001378923654556, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.013062526797875762, + "grad_norm": 0.08663681894540787, + "kept_token_count": 572.6875, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9836047530174256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.40440484858118e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8779167175293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.88511352539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25040, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30014644265174867, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.01038189404644072, + "grad_norm": 0.07497833669185638, + "kept_token_count": 572.5265625, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9838171482086182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.321445325738751e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.52253112792968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.47528381347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25050, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.2999686986207962, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.012977189384400845, + "grad_norm": 0.0818752869963646, + "kept_token_count": 572.60625, + "learning_rate": 2.170392717566534e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9828647494316101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.693480063404422e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42446899414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3318206787109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25060, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.29915170669555663, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.010884013911709189, + "grad_norm": 0.12454300373792648, + "kept_token_count": 573.3984375, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9819947719573975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.033312613202725e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.16224899291993, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4591430664062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25070, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.29901248812675474, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.014101951709017157, + "grad_norm": 0.0790623128414154, + "kept_token_count": 573.5, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.983249843120575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.543049275933299e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.03007736206055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0539276123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.2994719535112381, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010769808152690529, + "grad_norm": 0.10459785163402557, + "kept_token_count": 573.278125, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9826490938663482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.777718590456061e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.81747817993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1792877197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25090, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2997755497694016, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.010482810251414777, + "grad_norm": 0.10256287455558777, + "kept_token_count": 573.0109375, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9834717810153961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.456358314608224e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.22480545043945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4368560791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25100, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29984012246131897, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.012601039325818419, + "grad_norm": 0.09818078577518463, + "kept_token_count": 572.884375, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9833840370178223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.490628511528485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.49374313354492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5481201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25110, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.299869567155838, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.010074233449995518, + "grad_norm": 0.116374172270298, + "kept_token_count": 572.809375, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9833699762821198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.496124224213418e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.83229141235351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4692138671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25120, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.2996023058891296, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.012640557577833533, + "grad_norm": 0.113059863448143, + "kept_token_count": 573.0296875, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9828191518783569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.711288515361957e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27686614990235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4645294189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25130, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.29937519133090973, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.012297439947724342, + "grad_norm": 0.07252561300992966, + "kept_token_count": 573.0640625, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9821186244487763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.98492702213116e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.88022689819336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.42994079589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25140, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.2994385153055191, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.010574831906706095, + "grad_norm": 0.06362632662057877, + "kept_token_count": 573.3265625, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9825602471828461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.812419669586234e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.7859001159668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.16895446777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25150, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.30051605105400087, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.011089653801172971, + "grad_norm": 0.0827188640832901, + "kept_token_count": 572.665625, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9835760772228241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.415615389414598e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.93770751953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.18935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25160, + "valid_token_count": 818.696875 + }, + { + "actual_token_mask_ratio": 0.29965047538280487, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.010616596974432468, + "grad_norm": 0.11296396702528, + "kept_token_count": 572.871875, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9824311912059784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.862841473775916e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.06758880615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.856201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.3011681169271469, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.011770207947120071, + "grad_norm": 0.1318385899066925, + "kept_token_count": 571.9921875, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.982789796590805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.722751022607554e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.8853530883789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.12568359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25180, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.29942525625228883, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.0114192602224648, + "grad_norm": 0.08195450901985168, + "kept_token_count": 573.14375, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9835850358009338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.412113871192559e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.69626235961914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6842102050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25190, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.30001484751701357, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.011503480561077594, + "grad_norm": 0.06364691257476807, + "kept_token_count": 572.8921875, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9826761782169342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.767136474081781e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.02733535766602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.34706726074216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25200, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2995057314634323, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.012102177366614342, + "grad_norm": 0.07019445300102234, + "kept_token_count": 573.1578125, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9823518216609954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.893837708048523e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.21420745849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3917236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25210, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3003664672374725, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.011412329226732253, + "grad_norm": 0.14949962496757507, + "kept_token_count": 572.3109375, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9837372899055481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.35264499578625e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.27440490722657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.35333251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25220, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.29959953725337984, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.0112469045445323, + "grad_norm": 0.06384015083312988, + "kept_token_count": 572.9046875, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9827054023742676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.755719368811697e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.02380294799805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6395782470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25230, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.299726402759552, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.010456446511670946, + "grad_norm": 0.06009313091635704, + "kept_token_count": 572.9015625, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9836546063423157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.384939806594047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90125122070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7731872558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25240, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.3000909000635147, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.011968113807961344, + "grad_norm": 0.08915914595127106, + "kept_token_count": 572.7, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9826303899288178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.785015793866477e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.53348007202149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6172180175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25250, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.300024151802063, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.01148170824162662, + "grad_norm": 0.08941077440977097, + "kept_token_count": 572.553125, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.982675838470459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.767267186660319e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.18149032592774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66485290527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25260, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2997280299663544, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.012901711650192738, + "grad_norm": 0.11507885903120041, + "kept_token_count": 573.1359375, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9831856667995453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.568113130924757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75092926025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1360260009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25270, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.2999337941408157, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.010600524814799428, + "grad_norm": 0.07271382212638855, + "kept_token_count": 572.9859375, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9828869819641113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.684787367703394e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.30456314086913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9223114013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25280, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.30041787922382357, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.010650331806391477, + "grad_norm": 0.0626116618514061, + "kept_token_count": 572.1171875, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9833489060401917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.504352058982477e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.29567489624023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5337646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25290, + "valid_token_count": 817.7984375 + }, + { + "actual_token_mask_ratio": 0.30087795853614807, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.010675303870812058, + "grad_norm": 0.06673980504274368, + "kept_token_count": 572.009375, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9825256764888763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.825926502642687e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.50491638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0864776611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3002220571041107, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.011584890121594072, + "grad_norm": 0.09071376919746399, + "kept_token_count": 572.8515625, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9829227149486541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.670833899988793e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.14748458862304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.57575073242185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25310, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.3002924174070358, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.010301308706402779, + "grad_norm": 0.0749330073595047, + "kept_token_count": 572.703125, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9842167258262634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.165363483887632e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.66760559082032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8779052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25320, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.2995122462511063, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.011998393526300787, + "grad_norm": 0.11082839965820312, + "kept_token_count": 573.334375, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9823824107646942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.881884000904392e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.23320999145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.52147216796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25330, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.29928843677043915, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.011798259150236846, + "grad_norm": 0.11378058046102524, + "kept_token_count": 573.0734375, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9827543497085571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.73659989843145e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.59333572387695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8299255371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25340, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.29968683421611786, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012075905920937657, + "grad_norm": 0.07550480216741562, + "kept_token_count": 572.9921875, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9836293816566467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.394793672370724e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.2391860961914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6708190917969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2990826040506363, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.01040620799176395, + "grad_norm": 0.07546071708202362, + "kept_token_count": 573.6078125, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9824977934360504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.836823959019967e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.65899429321288, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2053924560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25360, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2995943367481232, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.011251967446878553, + "grad_norm": 0.1188071072101593, + "kept_token_count": 573.1109375, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9830302059650421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.628845731029286e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.87996520996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.056689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25370, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29946679472923277, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.011377469683066011, + "grad_norm": 0.09022187441587448, + "kept_token_count": 573.234375, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9828839302062988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.685982625640464e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.22735977172852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4769653320312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25380, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2996728539466858, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.011526668164879083, + "grad_norm": 0.10293863713741302, + "kept_token_count": 572.965625, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9819364905357361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05607708368916e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.00856246948243, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39646911621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25390, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2996057689189911, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.009948127064853906, + "grad_norm": 0.0956040695309639, + "kept_token_count": 573.1484375, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9829336822032928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.666547124041244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90691909790038, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2808776855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25400, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.3002683937549591, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.012175948452204466, + "grad_norm": 0.10098607838153839, + "kept_token_count": 572.559375, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.982650750875473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.777074995625298e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.21899490356445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.85639038085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25410, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.2993344247341156, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.012842506356537341, + "grad_norm": 0.08417969197034836, + "kept_token_count": 573.596875, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9836459398269654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.38832505501341e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.39852905273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0332946777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25420, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.2997743010520935, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.010869639227166772, + "grad_norm": 0.07500480860471725, + "kept_token_count": 573.0203125, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9833771228790283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.493335204140748e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.00656356811524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8829345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25430, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.299881175160408, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.009880912303924561, + "grad_norm": 0.08782397210597992, + "kept_token_count": 572.8015625, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9824318826198578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.862565351184458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.87984008789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6925018310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25440, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.2990823358297348, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.011881989380344748, + "grad_norm": 0.08879996836185455, + "kept_token_count": 573.853125, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9840475022792816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.231457336980384e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.7397087097168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9538513183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25450, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.30024600923061373, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.009430451830849052, + "grad_norm": 0.1014789491891861, + "kept_token_count": 572.5828125, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.9836551249027252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.384737571352161e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11694717407227, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.19598693847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25460, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29966323673725126, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.012203391874209046, + "grad_norm": 0.08551421016454697, + "kept_token_count": 573.034375, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9824963510036469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.837380169599783e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.56933059692383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.87721252441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25470, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3007757991552353, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.0127952235750854, + "grad_norm": 0.08276678621768951, + "kept_token_count": 572.0515625, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9824227094650269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.866143085062503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.12355117797851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.44396057128904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25480, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.30027690529823303, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.01123372595757246, + "grad_norm": 0.06456004083156586, + "kept_token_count": 572.3609375, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9832175076007843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.555682411999442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.93788528442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.64231872558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25490, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2997992366552353, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.012376107834279537, + "grad_norm": 0.1208776906132698, + "kept_token_count": 573.0875, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9832493364810944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.543244708154817e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.65542984008789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.86179504394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.30003432631492616, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.011163783865049482, + "grad_norm": 0.11242851614952087, + "kept_token_count": 572.8234375, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.982957535982132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.657228695985395e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.05944442749023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3497833251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25510, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.29873342514038087, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.011314598890021444, + "grad_norm": 0.07367958873510361, + "kept_token_count": 574.04375, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9828073978424072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.715879499097355e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.54033889770508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.46019592285154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25520, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.2995142102241516, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.012771346187219024, + "grad_norm": 0.09344872087240219, + "kept_token_count": 573.1578125, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9833909392356872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.487931459560059e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.69550170898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.75290832519534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25530, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.3009074926376343, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.01140166143886745, + "grad_norm": 0.19880209863185883, + "kept_token_count": 572.1, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9836981296539307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.367940404743422e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.94953842163086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6881042480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25540, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.30008555948734283, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012022278131917119, + "grad_norm": 0.09054704010486603, + "kept_token_count": 572.7, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9826777517795563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.766520891687833e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.80120544433593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.91761779785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25550, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.29831494092941285, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.012277399376034736, + "grad_norm": 0.12099549174308777, + "kept_token_count": 574.3203125, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9833150982856751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.517563633678947e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.33631286621093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.147998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25560, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3002719521522522, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.012470916006714105, + "grad_norm": 0.09939668327569962, + "kept_token_count": 572.528125, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9833092212677002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.519856688100845e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.56532363891601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5648162841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25570, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2999588161706924, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.011579529708251357, + "grad_norm": 0.13766999542713165, + "kept_token_count": 572.9734375, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9828999400138855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.679728721792344e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.2679557800293, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.67095947265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25580, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.30001001060009, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.012931601144373417, + "grad_norm": 0.09653575718402863, + "kept_token_count": 572.8015625, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9826920986175537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.760912074241788e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22496795654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.01720275878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25590, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.30019529163837433, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.012219373555853962, + "grad_norm": 0.10401441901922226, + "kept_token_count": 572.8671875, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9825002491474152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.83585982187651e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.56335830688477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5429962158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25600, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.300182381272316, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.012163442466408015, + "grad_norm": 0.10477260500192642, + "kept_token_count": 572.4828125, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9823408782482147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.898114370414987e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.93020477294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.60487670898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25610, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2997505396604538, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.010429737530648709, + "grad_norm": 0.11936406046152115, + "kept_token_count": 573.084375, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9831399559974671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.585972805623896e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52235488891601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.476416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3005108207464218, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.012049707397818566, + "grad_norm": 0.14389567077159882, + "kept_token_count": 572.5703125, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9828854560852051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.685383814328816e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.97979888916015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.38502807617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25630, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.2987093687057495, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.010076310345903039, + "grad_norm": 0.07658365368843079, + "kept_token_count": 573.64375, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9831049323081971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.59965124214068e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.54756927490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1457214355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25640, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.3007829010486603, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.011961797252297401, + "grad_norm": 0.11854472011327744, + "kept_token_count": 572.128125, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9822129607200623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.948080699658022e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.73536376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4427856445312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25650, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2999538600444794, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.009914295002818107, + "grad_norm": 0.08302297443151474, + "kept_token_count": 572.8953125, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.982133811712265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.978996098041534e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20548477172852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4052673339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25660, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3003880769014359, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.011561763379722834, + "grad_norm": 0.12896627187728882, + "kept_token_count": 572.5203125, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9838956415653228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.290785131568555e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.19519577026367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8414611816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25670, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3000035285949707, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.013103364454582333, + "grad_norm": 0.11450980603694916, + "kept_token_count": 572.5953125, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9826688289642334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.770010877517052e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3287727355957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.00913391113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25680, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.3001675307750702, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.011733958777040244, + "grad_norm": 0.11720132827758789, + "kept_token_count": 572.85, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.982911103963852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.675369222648442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.7985237121582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0430969238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25690, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.29902719557285307, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.01141988462768495, + "grad_norm": 0.07158960402011871, + "kept_token_count": 573.5171875, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9832245707511902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.552919876412489e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.720556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3199493408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25700, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29961795508861544, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.012057945877313615, + "grad_norm": 0.10082995891571045, + "kept_token_count": 572.8390625, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9830760598182678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.610935924982187e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.15683517456054, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.21435546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25710, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.30034968852996824, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.01099313902668655, + "grad_norm": 0.08658869564533234, + "kept_token_count": 572.6703125, + "learning_rate": 1.639409017553048e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9835254549980164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.435393006540835e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.82480850219727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0361572265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25720, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.30072781443595886, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.010237078554928303, + "grad_norm": 0.07697216421365738, + "kept_token_count": 572.0953125, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9836188495159149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.39889964077156e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.33407363891601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2698455810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25730, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.29994064569473267, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.012525125779211521, + "grad_norm": 0.09729277342557907, + "kept_token_count": 572.6734375, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9832504570484162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.542809314851184e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.95705184936523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.96982421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25740, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.3001021176576614, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.010335208009928465, + "grad_norm": 0.08597645163536072, + "kept_token_count": 572.5734375, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9824655950069427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.849398596386891e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.54172897338867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8591369628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25750, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.30037702322006227, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.010879964753985404, + "grad_norm": 0.10813050717115402, + "kept_token_count": 572.5140625, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9828617572784424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.694646472169553e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.83480987548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.48345336914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25760, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3004310101270676, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.011337452428415417, + "grad_norm": 0.08319643139839172, + "kept_token_count": 572.41875, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.98282470703125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.709115550620481e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.29955825805663, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.09185485839845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3002501368522644, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.010680429404601455, + "grad_norm": 0.09403091669082642, + "kept_token_count": 572.51875, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9832828998565674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.530138889502269e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.25844116210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9774536132812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25780, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29991725385189055, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.013528030831366777, + "grad_norm": 0.09675870090723038, + "kept_token_count": 572.75625, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9830652594566345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.615153179154732e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.13603286743164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.93443908691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25790, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.30052606463432313, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.011214252235367895, + "grad_norm": 0.07271583378314972, + "kept_token_count": 572.178125, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.983263123035431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.537866320286412e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.03953475952149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4602386474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25800, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2984116166830063, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.010706581175327301, + "grad_norm": 0.11263195425271988, + "kept_token_count": 574.1328125, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9831369876861572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.587128445971757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.58490676879883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.38580017089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25810, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3005305647850037, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.011204885924234987, + "grad_norm": 0.08051806688308716, + "kept_token_count": 572.296875, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9830747365951538, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.611451281059998e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.30956268310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5890777587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25820, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.3000668495893478, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.010619367891922592, + "grad_norm": 0.055064231157302856, + "kept_token_count": 572.8609375, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9821931481361389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.955816097615752e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.42914276123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.07146911621095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25830, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.2997634172439575, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.012110749259591103, + "grad_norm": 0.15530724823474884, + "kept_token_count": 573.0328125, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.98309485912323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.603588699363172e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.40119018554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3778900146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25840, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2999128520488739, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.010701631475239993, + "grad_norm": 0.08208677172660828, + "kept_token_count": 573.071875, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9816012263298035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.187044539023191e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.52944946289062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.424365234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25850, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.2996653258800507, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.012279450055211782, + "grad_norm": 0.07597912102937698, + "kept_token_count": 573.0484375, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9831042528152466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.599923799512907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.10112762451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7693664550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25860, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30142697393894197, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.012176876422017813, + "grad_norm": 0.11166398227214813, + "kept_token_count": 571.8953125, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9820731997489929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.002673737588339e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.55148086547851, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0037506103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25870, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.29938185811042783, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.010465836059302092, + "grad_norm": 0.1101667582988739, + "kept_token_count": 573.296875, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9828120410442353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.714064402331132e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.87004699707032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4666717529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3001387983560562, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.01293584150262177, + "grad_norm": 0.10853882133960724, + "kept_token_count": 572.453125, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9834644258022308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.459228352468926e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.47804412841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7969696044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25890, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.3003935217857361, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.012817082041874529, + "grad_norm": 0.1230544000864029, + "kept_token_count": 572.375, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9823576927185058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.891544180689379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.63495254516602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.68906555175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25900, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.3007901251316071, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.010704547725617885, + "grad_norm": 0.07743168622255325, + "kept_token_count": 572.021875, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9831430375576019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.584770308109e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.89820404052735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9114074707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25910, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2994354099035263, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.010488530760630966, + "grad_norm": 0.13487985730171204, + "kept_token_count": 573.4515625, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9830752193927765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.611261487705633e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.90883941650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47060241699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25920, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.30001998543739317, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.010902413446456195, + "grad_norm": 0.07863258570432663, + "kept_token_count": 572.5515625, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9832867503166198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.528628800879232e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.74223251342774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.80958557128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25930, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.2993727207183838, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.011313942214474081, + "grad_norm": 0.08964166790246964, + "kept_token_count": 573.1484375, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.982975572347641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.650190334767103e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6640724182129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9937744140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25940, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.300490602850914, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.01120258872397244, + "grad_norm": 0.0932859405875206, + "kept_token_count": 572.36875, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9828846633434296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.685704138362781e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.39168243408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8931610107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25950, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2998965620994568, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.010963457729667425, + "grad_norm": 0.1342722475528717, + "kept_token_count": 573.096875, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9822355210781097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.939268751011695e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77516021728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3163177490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25960, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.3001083701848984, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.011368324747309089, + "grad_norm": 0.07818536460399628, + "kept_token_count": 572.865625, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9831965744495392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.563854221894872e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.43414688110352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9415344238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25970, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.3001655042171478, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.013500412460416555, + "grad_norm": 0.08705858141183853, + "kept_token_count": 572.7828125, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9822848975658417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.919982988620176e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.35564270019532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.90610961914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25980, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.30085475742816925, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.013038161769509315, + "grad_norm": 0.07325849682092667, + "kept_token_count": 572.2578125, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9823297977447509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.902442983118817e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.39757385253907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8886474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25990, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.3001772195100784, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.010684676328673958, + "grad_norm": 0.08818121254444122, + "kept_token_count": 572.4921875, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9831446766853332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.584129041584675e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.70228881835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6856201171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26000, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.2987053722143173, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.01303126672282815, + "grad_norm": 0.10935333371162415, + "kept_token_count": 573.7875, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9819459617137909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.05237365764333e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.00238647460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.87261657714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2995510071516037, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.011587506625801325, + "grad_norm": 0.16509957611560822, + "kept_token_count": 573.109375, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.981556236743927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.204612884379458e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.4443458557129, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3726287841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.299067485332489, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.010952209867537022, + "grad_norm": 0.12602779269218445, + "kept_token_count": 573.865625, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9823367178440094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.899734726175665e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.29660415649414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5461486816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26030, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.30022440254688265, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.012520858086645603, + "grad_norm": 0.09802214801311493, + "kept_token_count": 572.3296875, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9822957634925842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.915733465575613e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.10840148925782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7043121337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26040, + "valid_token_count": 817.8765625 + }, + { + "actual_token_mask_ratio": 0.30005963146686554, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.010568463569507002, + "grad_norm": 0.07789832353591919, + "kept_token_count": 572.7, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9828768908977509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.688727953587658e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.84963607788086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.89247131347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26050, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.3002334475517273, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.013307439256459474, + "grad_norm": 0.0993761271238327, + "kept_token_count": 572.5140625, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9838028430938721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.327031005639582e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.81779327392579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0912261962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26060, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2995771527290344, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.010343661252409219, + "grad_norm": 0.09136592596769333, + "kept_token_count": 573.3703125, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9834130644798279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.479290132119787e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.63853225708007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7436004638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26070, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.300027871131897, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.011012309556826949, + "grad_norm": 0.07444291561841965, + "kept_token_count": 572.4796875, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9829427301883698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66301082674181e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.06210632324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0822021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26080, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.299601736664772, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.009389846492558718, + "grad_norm": 0.08817584067583084, + "kept_token_count": 573.171875, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0094, + "mgd_cos_sim": 0.9827919006347656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.721932768414263e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.23246688842774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.71216735839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26090, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2994398772716522, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.01094800466671586, + "grad_norm": 0.09681957215070724, + "kept_token_count": 573.41875, + "learning_rate": 1.365591193327712e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9834176003932953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.477517817984335e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.23124618530274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41172790527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26100, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.3020416796207428, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.013448426173999906, + "grad_norm": 0.07968463003635406, + "kept_token_count": 571.165625, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9829878449440003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.645390967605636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.46333236694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1712219238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26110, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2999970048666, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.012739472929388285, + "grad_norm": 0.11832768470048904, + "kept_token_count": 572.7546875, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9827264904975891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.747484912921209e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.51898498535157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.12945861816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26120, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30035941302776337, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.011001532012596727, + "grad_norm": 0.10593684762716293, + "kept_token_count": 572.3921875, + "learning_rate": 1.344986725812959e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9822946012020111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.916190832271241e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.5812484741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.93312072753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26130, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.299002343416214, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.011127006728202105, + "grad_norm": 0.09994002431631088, + "kept_token_count": 573.5390625, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.982305932044983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.911767995916307e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.88155975341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6616241455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30037380158901217, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.011635560309514403, + "grad_norm": 0.09403645247220993, + "kept_token_count": 572.46875, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9829011619091034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.679258949588984e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.95170822143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.21300354003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26150, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30018003582954406, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.01115490118972957, + "grad_norm": 0.104075126349926, + "kept_token_count": 572.3828125, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9818806767463684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.077881055010949e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96997146606445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2604217529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26160, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.29927290976047516, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.01135740727186203, + "grad_norm": 0.11311950534582138, + "kept_token_count": 573.64375, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9834043741226196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.48268272925634e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.43688201904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.855126953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26170, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.29963179528713224, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.01043992517516017, + "grad_norm": 0.09634998440742493, + "kept_token_count": 572.8859375, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9828394711017608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.703349645249545e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.13683166503907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4270263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26180, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.2996372014284134, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.011371153825893997, + "grad_norm": 0.10015159100294113, + "kept_token_count": 572.75, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9821270644664765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.981629121582955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.52592239379882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.34446716308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26190, + "valid_token_count": 817.7875 + }, + { + "actual_token_mask_ratio": 0.29915258586406707, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.011425551027059555, + "grad_norm": 0.1562364399433136, + "kept_token_count": 573.65, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9822553336620331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.931529678695369e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.50915603637695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.23304443359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26200, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.30042955577373504, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.010302762174978853, + "grad_norm": 0.0868087112903595, + "kept_token_count": 572.5171875, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9833800852298736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.492172906291671e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.62798461914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.21022338867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26210, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3001750648021698, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.01194666586816311, + "grad_norm": 0.09634966403245926, + "kept_token_count": 572.7203125, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9828976273536683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.680631631752476e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.20610198974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7353088378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26220, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.300354540348053, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.011997877433896065, + "grad_norm": 0.08859790116548538, + "kept_token_count": 572.4484375, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9823280870914459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.903108915139456e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.07244338989258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7378356933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.30015057027339936, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.010889557097107172, + "grad_norm": 0.10429876297712326, + "kept_token_count": 572.496875, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9838181912899018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.321036125882529e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.98448104858399, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.83591613769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26240, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.30056641101837156, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.009806715184822679, + "grad_norm": 0.07279577106237411, + "kept_token_count": 572.3484375, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.983190244436264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.566329102497548e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7411605834961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6963653564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26250, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.29974383413791655, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.014229374891147017, + "grad_norm": 0.0852954313158989, + "kept_token_count": 572.9921875, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9829345583915711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.666208828391973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.02242050170898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.99705200195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26260, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30048059225082396, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.00995746711269021, + "grad_norm": 0.06886923313140869, + "kept_token_count": 572.44375, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9831333994865418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.588537144125439e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.07298126220704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3284210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29963064193725586, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.011943415878340602, + "grad_norm": 0.08149777352809906, + "kept_token_count": 573.15, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9823062062263489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.91165536409244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.52109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5005157470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26280, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30002541542053224, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.010869052028283477, + "grad_norm": 0.0909455418586731, + "kept_token_count": 572.4578125, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9830304384231567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.628755108977203e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.35909576416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1881378173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26290, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.29965962171554567, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.009502516686916351, + "grad_norm": 0.10030537098646164, + "kept_token_count": 573.4171875, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.9834076821804046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.481395139417146e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.1500343322754, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.59560546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26300, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.29937055706977844, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.014027383271604776, + "grad_norm": 0.10010562837123871, + "kept_token_count": 573.5296875, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9828093230724335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.71512374537997e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.37083740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26310, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.2997538834810257, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.010559654468670488, + "grad_norm": 0.10192914307117462, + "kept_token_count": 572.990625, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9820589661598206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.008234897512011e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3691619873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3061187744141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26320, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30017160475254057, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.012447378225624561, + "grad_norm": 0.06303852051496506, + "kept_token_count": 573.0703125, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9837502181529999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.347589587676338e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.71852416992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.83339538574216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26330, + "valid_token_count": 818.8765625 + }, + { + "actual_token_mask_ratio": 0.30061601996421816, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.015096276067197323, + "grad_norm": 0.10454558581113815, + "kept_token_count": 572.3078125, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9823688626289367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.887179479235783e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.04308700561523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4918518066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26340, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.30076801478862764, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.01142700295895338, + "grad_norm": 0.09565582871437073, + "kept_token_count": 571.9640625, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9830739796161652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.611738208448515e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.47743911743164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7265197753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26350, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.2995796799659729, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.009628728637471795, + "grad_norm": 0.08959213644266129, + "kept_token_count": 573.0390625, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9822363138198853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.938963051652535e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.99996185302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6091827392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26360, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.30040326416492463, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.011474596010521055, + "grad_norm": 0.09894196689128876, + "kept_token_count": 572.5453125, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.983364599943161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.49822090053931e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.36746063232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6813568115234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26370, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29992777705192564, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.012608121149241924, + "grad_norm": 0.11387582123279572, + "kept_token_count": 572.9984375, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9825317442417145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.82355894241482e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.19443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0479278564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26380, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.2993540406227112, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.013449735008180141, + "grad_norm": 0.10156039148569107, + "kept_token_count": 573.0828125, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9830001771450043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.640572464675643e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.65079498291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0727081298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26390, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.29971722662448885, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.012532517872750759, + "grad_norm": 0.09272110462188721, + "kept_token_count": 573.1375, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9824458122253418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.857125044916757e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.20856094360352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4674133300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26400, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.3006139934062958, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.009685853030532598, + "grad_norm": 0.07950974255800247, + "kept_token_count": 572.5078125, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.982839047908783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.703515718982089e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.75175628662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5214782714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.3013095766305923, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.010522643709555269, + "grad_norm": 0.08964240550994873, + "kept_token_count": 571.7203125, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9826497793197632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.777447124477476e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.5613006591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2152038574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26420, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29967654347419737, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.012362853763625026, + "grad_norm": 0.11418763548135757, + "kept_token_count": 572.9640625, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9822661280632019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.927309550519567e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.14594421386718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9265380859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26430, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.3003510683774948, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.011122245574370027, + "grad_norm": 0.08294003456830978, + "kept_token_count": 572.6734375, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9840211570262909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.24176231212914e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.31254806518555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7350372314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26440, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2996995598077774, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.009826120175421237, + "grad_norm": 0.1052493304014206, + "kept_token_count": 573.078125, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9831567823886871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.579400251212064e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.47764358520507, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4243499755859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26450, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2996633261442184, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.010033143730834127, + "grad_norm": 0.06864707916975021, + "kept_token_count": 573.1171875, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9828710734844208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.691004819003865e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93423538208008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6218017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26460, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.299453067779541, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.011713787214830518, + "grad_norm": 0.11232815682888031, + "kept_token_count": 573.215625, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9834171891212463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.477674942289014e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.90031356811524, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.76145935058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26470, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.29877387285232543, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.011792494915425777, + "grad_norm": 0.09473826736211777, + "kept_token_count": 573.69375, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9828753054141999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.689351503155194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.60912017822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5760833740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26480, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.3009448707103729, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.011028898227959872, + "grad_norm": 0.1027999073266983, + "kept_token_count": 572.084375, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.983749943971634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.347692360577639e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.58294830322265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.91029357910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26490, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3000321894884109, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.011484979325905443, + "grad_norm": 0.07302680611610413, + "kept_token_count": 572.63125, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9835510432720185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.425387291528751e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.30915832519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.32317504882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26500, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.300457289814949, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.01219191295094788, + "grad_norm": 0.08494684100151062, + "kept_token_count": 572.5015625, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9827479302883149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.739111413480713e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.48102188110352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39052429199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26510, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2998517036437988, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.009842659207060933, + "grad_norm": 0.10537488758563995, + "kept_token_count": 573.10625, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9834467113018036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.466150880442e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.1157829284668, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2287261962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26520, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.299778214097023, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.012254803813993931, + "grad_norm": 0.07169045507907867, + "kept_token_count": 572.7515625, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9836801826953888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.374949371092952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.52994003295899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4197174072266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26530, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.2992642283439636, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.010843648901209235, + "grad_norm": 0.07333975285291672, + "kept_token_count": 573.2703125, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9823659062385559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.888340212753973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.68741302490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8495727539063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26540, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2998350590467453, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.01136036468669772, + "grad_norm": 0.09657842665910721, + "kept_token_count": 572.971875, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9835835874080658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.412677685148082e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.33346252441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4988159179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26550, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30027044117450713, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.011133662750944495, + "grad_norm": 0.09998498857021332, + "kept_token_count": 572.584375, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9826995730400085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.7579944152385e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.77514953613282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.87591247558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26560, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.2997400611639023, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.015275588445365428, + "grad_norm": 0.08342117071151733, + "kept_token_count": 572.9453125, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9827984511852265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.719378579873591e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.75570983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9611419677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26570, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3007518589496613, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.011864142958074808, + "grad_norm": 0.10079578310251236, + "kept_token_count": 572.1359375, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9843771517276764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.1027001356706025e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.85631713867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4111022949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26580, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.29881349205970764, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.01025808583945036, + "grad_norm": 0.0863020047545433, + "kept_token_count": 573.6890625, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.982596492767334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.79826749546919e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.81843719482421, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.77288208007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26590, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.30048870742321016, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.013751681288704276, + "grad_norm": 0.10466960817575455, + "kept_token_count": 572.325, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9824954092502594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.837747750978451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.3855194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.363427734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26600, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.29990683794021605, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.011548070842400194, + "grad_norm": 0.08866292238235474, + "kept_token_count": 572.809375, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9833062708377838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.521014729514718e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.34361267089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7140228271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26610, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3001010924577713, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.00986173739656806, + "grad_norm": 0.10520169883966446, + "kept_token_count": 572.671875, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9828072249889374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.71594294544775e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.81108169555664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.133056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26620, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30006812512874603, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.012045417819172144, + "grad_norm": 0.1192193552851677, + "kept_token_count": 572.703125, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9828611195087433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.694893672829494e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.01404418945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8629669189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26630, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3008278399705887, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.011991487024351955, + "grad_norm": 0.057292696088552475, + "kept_token_count": 572.2625, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9833737552165985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.494646186183672e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.29750061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8900970458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26640, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.3011531174182892, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.010877486597746611, + "grad_norm": 0.13397392630577087, + "kept_token_count": 571.859375, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9832198202610016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.554776300617959e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38098220825195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4881164550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.29988894164562224, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.011286428710445762, + "grad_norm": 0.0689648762345314, + "kept_token_count": 572.8265625, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9829729676246644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.651195508311503e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.35773162841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.72740173339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26660, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.30148965716362, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.011044480977579952, + "grad_norm": 0.0601980984210968, + "kept_token_count": 571.6, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9827214896678924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.74944145430345e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.79742431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6168975830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26670, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29972451329231264, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.010705021116882563, + "grad_norm": 0.09513126313686371, + "kept_token_count": 573.0171875, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9831899881362915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.566434094565921e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.69915237426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9607727050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26680, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3001786768436432, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.011201335955411196, + "grad_norm": 0.08744959533214569, + "kept_token_count": 572.7765625, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9829688370227814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.652817319263704e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.71815948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1120941162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26690, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2995542913675308, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.012284639570862056, + "grad_norm": 0.07871348410844803, + "kept_token_count": 573.265625, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9829370737075805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.665221371804364e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.75228805541992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.98634643554686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26700, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2994563102722168, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.011926529882475733, + "grad_norm": 0.08365533500909805, + "kept_token_count": 573.3890625, + "learning_rate": 9.762452810644068e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9835934460163116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.408826848200989e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.30987701416015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.69724426269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26710, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.3004247725009918, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.012454409711062908, + "grad_norm": 0.07809620350599289, + "kept_token_count": 572.3171875, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9829618394374847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.655546094407328e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.32056503295898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9217498779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.29877821207046507, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.014868734683841468, + "grad_norm": 0.11648353189229965, + "kept_token_count": 573.7515625, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9824279606342315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.864093957119622e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.76760635375976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3708801269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26730, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2998855859041214, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.011387332202866673, + "grad_norm": 0.07696349173784256, + "kept_token_count": 572.6875, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9823897540569305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.879015636513941e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.36613693237305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6822540283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26740, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.30115489959716796, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.010122589021921157, + "grad_norm": 0.06729342043399811, + "kept_token_count": 572.0015625, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9824094533920288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.871327314001973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5218734741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8560516357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26750, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.2998680740594864, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.012679409934207797, + "grad_norm": 0.08679617196321487, + "kept_token_count": 573.0296875, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9821439445018768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.975040378165431e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.51302795410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9644714355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26760, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.3000830203294754, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.011559011228382588, + "grad_norm": 0.05759287253022194, + "kept_token_count": 572.6171875, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9829654157161712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.654148310190067e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5020133972168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.99347229003905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26770, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.29980539083480834, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.010650344053283333, + "grad_norm": 0.06968798488378525, + "kept_token_count": 573.090625, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9827132403850556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.752659646735992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.14108505249024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.87437133789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26780, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.30134322345256803, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.013413937389850616, + "grad_norm": 0.13874506950378418, + "kept_token_count": 571.728125, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9821448922157288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.974663410801441e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90989837646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.61753540039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30021830797195437, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.011913542728871107, + "grad_norm": 0.09620293974876404, + "kept_token_count": 572.58125, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9826038599014282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.795388108002954e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08426513671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5033721923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26800, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29946134686470033, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.010197664145380259, + "grad_norm": 0.09188932925462723, + "kept_token_count": 573.2390625, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9825069427490234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.833239676780067e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.62462692260742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6216552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29913392663002014, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.010511066112667323, + "grad_norm": 0.11730124801397324, + "kept_token_count": 573.43125, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9823417842388154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.89776221406646e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.54375534057617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5635925292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26820, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30018593966960905, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.011773896729573607, + "grad_norm": 0.05373496934771538, + "kept_token_count": 572.778125, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9828537166118622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.697781864204445e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.30865325927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.83526306152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26830, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.299835079908371, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.009799117781221866, + "grad_norm": 0.0759883001446724, + "kept_token_count": 572.9171875, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9835903882980347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.410023124772123e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.8464141845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.39625244140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26840, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.30080671012401583, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.011859353305771947, + "grad_norm": 0.10443881154060364, + "kept_token_count": 572.178125, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9818102955818176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.105370968929492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.76405410766601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.01015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26850, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2999630063772202, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.011580336559563875, + "grad_norm": 0.09781696647405624, + "kept_token_count": 572.7890625, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9830275535583496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.629877680097707e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.6175323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0831268310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29972912967205045, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.010673439409583808, + "grad_norm": 0.07652245461940765, + "kept_token_count": 572.9484375, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9831162095069885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.595241429749877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.85759658813477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3599914550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26870, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30064430832862854, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.010799478180706501, + "grad_norm": 0.09395620971918106, + "kept_token_count": 572.471875, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9838524222373962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.307672374532558e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.95425872802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4872650146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26880, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.2991469830274582, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.010615988774225116, + "grad_norm": 0.07368044555187225, + "kept_token_count": 573.3796875, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9827387511730195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.742700243194122e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.24191207885742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.71861267089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26890, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.30145970582962034, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.01079483749344945, + "grad_norm": 0.07452470809221268, + "kept_token_count": 571.590625, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9831706523895264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.573979371751192e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.68061447143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3414794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26900, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3006511449813843, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.010383659973740578, + "grad_norm": 0.09588147699832916, + "kept_token_count": 572.3109375, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9823144316673279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.90844317432493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17386398315429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.87973937988284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26910, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30147054195404055, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.009918315522372722, + "grad_norm": 0.10430784523487091, + "kept_token_count": 571.5515625, + "learning_rate": 8.567936562219547e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9831186950206756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.594277147087269e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8434814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.45738525390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26920, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3003611922264099, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.011534706270322204, + "grad_norm": 0.09776362776756287, + "kept_token_count": 572.459375, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9834057867527009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.482126991613768e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.52685241699218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2181915283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26930, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30026601254940033, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.009244206361472607, + "grad_norm": 0.06473463773727417, + "kept_token_count": 572.5859375, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0093, + "mgd_cos_sim": 0.9830107927322388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.636427751800511e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.63892974853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4041687011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26940, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.30037344694137574, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.011267173755913972, + "grad_norm": 0.08544378727674484, + "kept_token_count": 572.4390625, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.983387565612793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.489249062724411e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.63468704223632, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4653717041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26950, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29909403920173644, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.011765818810090423, + "grad_norm": 0.1665852814912796, + "kept_token_count": 573.4546875, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9831304132938385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.58970184304053e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44333572387696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1880401611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26960, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.3002705365419388, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.008736724825575947, + "grad_norm": 0.09741126000881195, + "kept_token_count": 572.434375, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0088, + "mgd_cos_sim": 0.9836293518543243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.394803131115623e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.40428771972657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1645904541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26970, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.2994308203458786, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.011795631470158696, + "grad_norm": 0.08769340813159943, + "kept_token_count": 573.290625, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9835466027259827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.427127664210275e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.17440719604492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.49493103027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26980, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.300006577372551, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.011472101416438818, + "grad_norm": 0.09969604015350342, + "kept_token_count": 572.65625, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9826155602931976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.790818952140398e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.67437896728515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0716888427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26990, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2995547354221344, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.011864384356886148, + "grad_norm": 0.09131763875484467, + "kept_token_count": 573.2640625, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9833225846290589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.514641900139395e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.25128555297852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.54454345703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27000, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.30048902332782745, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.010486789001151919, + "grad_norm": 0.08541583269834518, + "kept_token_count": 572.153125, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.984281575679779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.140023688203656e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.59500122070312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.21949462890626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27010, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.2984881639480591, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.01145570920780301, + "grad_norm": 0.06449486315250397, + "kept_token_count": 574.0125, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.983575040102005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.416016985895112e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2839485168457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.72645263671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27020, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2992364317178726, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.011938929371535779, + "grad_norm": 0.07588879764080048, + "kept_token_count": 573.475, + "learning_rate": 7.972443479549035e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9830916464328766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.604842710657976e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17726593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1231292724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27030, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29984239041805266, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.011950170900672675, + "grad_norm": 0.07623997330665588, + "kept_token_count": 573.1203125, + "learning_rate": 7.919341901912236e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9832870244979859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.5285200253129e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.61940689086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7205474853516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27040, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.3009404480457306, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.011396003561094404, + "grad_norm": 0.09456906467676163, + "kept_token_count": 572.2421875, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9832408905029297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.546550503117032e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.02763519287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7866607666016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27050, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.2998184025287628, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.011420962121337652, + "grad_norm": 0.08850926905870438, + "kept_token_count": 573.1734375, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9833401083946228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.507791513286066e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.42858810424805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7235504150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27060, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.299489238858223, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.013537079840898514, + "grad_norm": 0.10058305412530899, + "kept_token_count": 573.1953125, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9823863863945007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.880338187329472e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.49705581665039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.14325561523435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27070, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30074494481086733, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.013381997216492891, + "grad_norm": 0.08897349238395691, + "kept_token_count": 572.20625, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9824955463409424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.837693508714438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.70533142089843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7758422851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27080, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.3006617099046707, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.01009538802318275, + "grad_norm": 0.09056215733289719, + "kept_token_count": 572.259375, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9834213554859161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.47604978439631e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.95704040527343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.330078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.29990264773368835, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.012094628252089023, + "grad_norm": 0.09709085524082184, + "kept_token_count": 572.7328125, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9834397435188293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.46886466711294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.62667236328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8329711914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27100, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30014235377311704, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.010273208282887936, + "grad_norm": 0.09356516599655151, + "kept_token_count": 572.7140625, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9831987261772156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.563013303093613e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.73232879638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9026336669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.29963645339012146, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.011100491369143128, + "grad_norm": 0.05424775183200836, + "kept_token_count": 573.075, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9834429323673248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.467618841270451e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.81151275634765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9011535644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27120, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29943837523460387, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.011970507260411978, + "grad_norm": 0.11605078727006912, + "kept_token_count": 573.55625, + "learning_rate": 7.449204239591345e-07, + "loss": 0.012, + "mgd_cos_sim": 0.98312548995018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.591620440303814e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.23551177978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.18027954101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27130, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.2990251868963242, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.010959918843582273, + "grad_norm": 0.09338553249835968, + "kept_token_count": 573.615625, + "learning_rate": 7.397831944839167e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9827765047550201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.727949330525007e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.97768325805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16416931152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27140, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.30139838457107543, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.010135202342644334, + "grad_norm": 0.08039814978837967, + "kept_token_count": 571.5125, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9825677156448365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.809502447140403e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.37882843017579, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.29026184082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27150, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.3003037631511688, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.009304976603016258, + "grad_norm": 0.08903408795595169, + "kept_token_count": 572.61875, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0093, + "mgd_cos_sim": 0.9828569650650024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.696517630189192e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.87632751464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.93418884277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27160, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30061708986759184, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.010540007753297687, + "grad_norm": 0.07457106560468674, + "kept_token_count": 572.4375, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9830658316612244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614929043280426e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.30716705322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.20501403808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27170, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.300138396024704, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.013484644144773484, + "grad_norm": 0.1391378939151764, + "kept_token_count": 572.5625, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9829476296901702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66110234305961e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.60327072143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.55181884765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27180, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.2999518871307373, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.011321275681257247, + "grad_norm": 0.0871744453907013, + "kept_token_count": 572.921875, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9831869840621948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.567605196323711e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.51043701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7095184326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27190, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30032373666763307, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.011913034319877624, + "grad_norm": 0.08097155392169952, + "kept_token_count": 572.6703125, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9831967294216156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.563798087881878e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.59690704345704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2223693847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27200, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.30000934600830076, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.01187912654131651, + "grad_norm": 0.107708640396595, + "kept_token_count": 572.9328125, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9826314151287079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.784626893931999e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.15093688964843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.60877685546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27210, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.29919747114181516, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.011139854835346342, + "grad_norm": 0.09719497710466385, + "kept_token_count": 573.4359375, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.983615243434906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.400310485332739e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.51693191528321, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.21946716308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.299247482419014, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.010291181318461896, + "grad_norm": 0.10179432481527328, + "kept_token_count": 573.471875, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9834435880184174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.467369676101952e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.07496185302735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.38421020507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27230, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2997592091560364, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.010899800341576338, + "grad_norm": 0.08486843854188919, + "kept_token_count": 572.9703125, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9829551696777343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.658153033640701e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.06714782714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0375610351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27240, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3002448409795761, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.011826793616637587, + "grad_norm": 0.08868369460105896, + "kept_token_count": 572.5375, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9834544956684113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.463108184107114e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.60209503173829, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0482238769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.29907186031341554, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.012461886182427407, + "grad_norm": 0.09983430802822113, + "kept_token_count": 573.5078125, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9830934762954712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.604134468943812e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90984954833985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2671752929688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27260, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29916426837444304, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.009985339036211372, + "grad_norm": 0.11473613977432251, + "kept_token_count": 573.434375, + "learning_rate": 6.745782579457993e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9832325458526612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.549809331772849e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.84712982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.78599853515624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27270, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.30042065382003785, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.011389768961817027, + "grad_norm": 0.07313153892755508, + "kept_token_count": 572.5171875, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9830319344997406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.628171431657392e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11052856445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5279968261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27280, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.30101293325424194, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.010824917862191797, + "grad_norm": 0.12319386005401611, + "kept_token_count": 571.7265625, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.983755350112915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.345586953102612e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.53491516113282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6749725341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27290, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.300945183634758, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.010831779893487692, + "grad_norm": 0.08716868609189987, + "kept_token_count": 571.990625, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9831950068473816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.564472678292078e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.94051818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.24501647949216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27300, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3001061290502548, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.01100743212737143, + "grad_norm": 0.07290450483560562, + "kept_token_count": 572.7671875, + "learning_rate": 6.551064140952417e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9834474980831146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.465836631832644e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20026321411133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.55491638183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27310, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2998194843530655, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.011436972115188838, + "grad_norm": 0.0963210016489029, + "kept_token_count": 573.053125, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9836119711399078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.401588143489789e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.31934356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9955535888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27320, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30003719627857206, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.011009185248985887, + "grad_norm": 0.09402786940336227, + "kept_token_count": 572.603125, + "learning_rate": 6.454750509527874e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9844040155410767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.092202529544011e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.4946174621582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7019500732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27330, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.30082018077373507, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.011636072024703025, + "grad_norm": 0.05022307485342026, + "kept_token_count": 572.0859375, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9825100779533387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.832018589193467e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.28911972045898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5328338623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27340, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29955872893333435, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.011216971697285772, + "grad_norm": 0.08380085974931717, + "kept_token_count": 573.1171875, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9837429404258728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.350430703605525e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.11326522827149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.96628723144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27350, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2992042899131775, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.010547271696850657, + "grad_norm": 0.07504761964082718, + "kept_token_count": 573.6359375, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9829174041748047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.672908275504597e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20405731201171, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4393707275391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27360, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.2994240641593933, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.009851635294035077, + "grad_norm": 0.06647476553916931, + "kept_token_count": 573.25625, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9829927206039428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.64348881400656e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.6896728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9910919189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27370, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.300136336684227, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.01087134233675897, + "grad_norm": 0.09957277029752731, + "kept_token_count": 572.553125, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9820648670196533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.00593096553348e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.16158294677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8002685546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27380, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.2999857097864151, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.009240366984158754, + "grad_norm": 0.04918402433395386, + "kept_token_count": 572.715625, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0093, + "mgd_cos_sim": 0.9825031042099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.834744417574256e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.97754821777343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5451324462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27390, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.30050640404224394, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.01123791905120015, + "grad_norm": 0.10578549653291702, + "kept_token_count": 572.375, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9834995329380035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.445513608923647e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11646347045898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5352844238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3003120839595795, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.011735772760584951, + "grad_norm": 0.10519377142190933, + "kept_token_count": 572.4328125, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9827703058719635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.730368368152994e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.39800720214843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4670654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27410, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3012784481048584, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.009742355300113558, + "grad_norm": 0.11365221440792084, + "kept_token_count": 571.43125, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9828953981399536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.68150452838745e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.9068115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48146362304686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27420, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.30108456015586854, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.01159321521408856, + "grad_norm": 0.08817992359399796, + "kept_token_count": 571.81875, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9831561923027039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.579631553904619e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.94364852905274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.71485290527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27430, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.3016828864812851, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.012136377766728402, + "grad_norm": 0.11498680710792542, + "kept_token_count": 571.3921875, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9838503360748291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.308480151346884e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.10719680786133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.62062072753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27440, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2993722438812256, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.012637355877086521, + "grad_norm": 0.09327644854784012, + "kept_token_count": 573.3234375, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9833018541336059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.522730036522262e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.61121520996093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16362915039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27450, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30021506547927856, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.012464649695903062, + "grad_norm": 0.09194241464138031, + "kept_token_count": 572.675, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9831349313259125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.587939788005315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.28593978881835, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2182983398437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27460, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3011357069015503, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.009793030889704823, + "grad_norm": 0.06326121836900711, + "kept_token_count": 571.7921875, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9835481941699982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.426514191844035e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.84225692749024, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3079498291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27470, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29992433488368986, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.010953673208132387, + "grad_norm": 0.10000618547201157, + "kept_token_count": 572.9515625, + "learning_rate": 5.75466871927543e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9832464098930359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.544386305904482e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.18891067504883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7386474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27480, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.2994126319885254, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.009978636330924929, + "grad_norm": 0.04345045983791351, + "kept_token_count": 573.0875, + "learning_rate": 5.709396685894463e-07, + "loss": 0.01, + "mgd_cos_sim": 0.984537523984909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.040055632183794e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.66245803833007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.48403015136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27490, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.3001034438610077, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.009961666446179152, + "grad_norm": 0.07890541851520538, + "kept_token_count": 572.7984375, + "learning_rate": 5.664299979202264e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9837906777858734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.331784388748928e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.65743026733398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8586120605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27500, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29992485642433164, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.010789772728458047, + "grad_norm": 0.083674855530262, + "kept_token_count": 572.625, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9824445903301239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.857593107270076e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.31154556274414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7144470214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27510, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.30088470578193666, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.012940626870840789, + "grad_norm": 0.10760680586099625, + "kept_token_count": 572.1484375, + "learning_rate": 5.574632764857974e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9842813909053802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.140102668723557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.85113754272462, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.00110778808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27520, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3003364443778992, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010103875352069736, + "grad_norm": 0.11488277465105057, + "kept_token_count": 572.6921875, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9824508666992188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.855154279037379e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07538986206055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5463500976563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27530, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.2994627714157104, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.011451804684475065, + "grad_norm": 0.08687016367912292, + "kept_token_count": 573.44375, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9833659708499909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.497685571957846e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.02207717895507, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.115576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27540, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.3004758805036545, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.010593395307660102, + "grad_norm": 0.06623531132936478, + "kept_token_count": 572.3796875, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9830714881420135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.612719371332788e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.03844985961913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0702789306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27550, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30102789402008057, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.012853843439370394, + "grad_norm": 0.0903339833021164, + "kept_token_count": 571.9109375, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.982829487323761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.707247084705159e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.16252365112305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4944549560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27560, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2993354082107544, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.01131546893157065, + "grad_norm": 0.12731145322322845, + "kept_token_count": 573.5375, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.983732157945633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.35464231891092e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.11535720825195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9936889648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27570, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.30068957805633545, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.009112663567066193, + "grad_norm": 0.08437727391719818, + "kept_token_count": 572.3828125, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0091, + "mgd_cos_sim": 0.9826551020145416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.775374713470228e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.98208160400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.54857177734374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27580, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.3000939339399338, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.010036093415692449, + "grad_norm": 0.05893850699067116, + "kept_token_count": 572.609375, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9835919559001922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.409408233594149e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.5438217163086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.87636108398436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27590, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.3010143369436264, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.011426384327933192, + "grad_norm": 0.10194917768239975, + "kept_token_count": 572.1375, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9839770019054412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.259004330786411e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.00496368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0517120361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27600, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.2994656145572662, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.00951678897254169, + "grad_norm": 0.0898001492023468, + "kept_token_count": 573.1734375, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9835858523845673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.41180118691409e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.08221130371093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3889190673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27610, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.3007392197847366, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.012237209593877196, + "grad_norm": 0.08061487227678299, + "kept_token_count": 572.421875, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9834004044532776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.484236655524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.42800750732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.53343505859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27620, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.3005084335803986, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.011819849349558353, + "grad_norm": 0.1610739678144455, + "kept_token_count": 572.525, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9834896981716156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.449346110457555e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.0801498413086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6265350341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27630, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.298917281627655, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.01300207651220262, + "grad_norm": 0.0821366086602211, + "kept_token_count": 573.740625, + "learning_rate": 5.051385685870924e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9830731868743896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.612053002754691e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91975250244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.24078063964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27640, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.2993696451187134, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.011062879534438252, + "grad_norm": 0.09448612481355667, + "kept_token_count": 573.2875, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9828834652900695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.686168926535174e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.88603057861329, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1117889404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27650, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2994157403707504, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.011354198679327965, + "grad_norm": 0.09768911451101303, + "kept_token_count": 573.4515625, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9828548312187195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.697351200273261e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.45323867797852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3614929199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27660, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.30035182237625124, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.010370805999264121, + "grad_norm": 0.08138485997915268, + "kept_token_count": 572.421875, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9826705515384674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.769332248950377e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.68777847290039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6454345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27670, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2985671371221542, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.01133846859447658, + "grad_norm": 0.10189105570316315, + "kept_token_count": 573.8171875, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9832776010036468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.532211264129729e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.29567947387696, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27680, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.30025061666965486, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.01352554871700704, + "grad_norm": 0.0742565467953682, + "kept_token_count": 572.7046875, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9820584237575531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.008448810665868e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.92988967895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1887603759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27690, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30022660493850706, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.009626922197639942, + "grad_norm": 0.0668991431593895, + "kept_token_count": 572.325, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9838460147380829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.310171957011335e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.44716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7387176513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27700, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.2995535105466843, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.012234430387616157, + "grad_norm": 0.10508354008197784, + "kept_token_count": 572.9484375, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9830773770809174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.610416603507474e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.4584358215332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.05372924804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27710, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.29936234652996063, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.014570037089288234, + "grad_norm": 0.1035952940583229, + "kept_token_count": 573.346875, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9827850878238678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.724595732521266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.30116500854493, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5456909179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2995230883359909, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.01173789780586958, + "grad_norm": 0.09668347984552383, + "kept_token_count": 572.98125, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9823424458503723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.897505445522257e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.02173767089843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0645233154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27730, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.3001884460449219, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.010787538997828961, + "grad_norm": 0.0831984132528305, + "kept_token_count": 572.521875, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9835222423076629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.436639741878026e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.08123321533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.70433654785154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.30044600963592527, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.0120462984777987, + "grad_norm": 0.06986568123102188, + "kept_token_count": 572.296875, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9825817108154297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.80404256854672e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.61794815063476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.63197937011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27750, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.29889391362667084, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.0114659926854074, + "grad_norm": 0.10810910165309906, + "kept_token_count": 573.790625, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9830139577388763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.635187855863478e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01227722167968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.54176025390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27760, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30004480481147766, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.00965032703243196, + "grad_norm": 0.0823739543557167, + "kept_token_count": 572.8078125, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.982827651500702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.707969987473916e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.46010131835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.93726501464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27770, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.3004464775323868, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.011677790386602282, + "grad_norm": 0.08742762356996536, + "kept_token_count": 572.4140625, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9836261510848999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.39605954347644e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.20093688964843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2126434326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27780, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2997880280017853, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.01173976636491716, + "grad_norm": 0.07210540771484375, + "kept_token_count": 573.06875, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9825656890869141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.810300765209831e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.44185638427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.46202087402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27790, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.30135211944580076, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.012417362630367279, + "grad_norm": 0.12690116465091705, + "kept_token_count": 571.615625, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9828721344470978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.690597292617895e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.43602142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92505798339846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27800, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.30005963742733, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.011399308359250427, + "grad_norm": 0.09650145471096039, + "kept_token_count": 572.6390625, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9830858826637268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.607094765058719e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20070037841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.79309997558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27810, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29920085072517394, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.012688728235661983, + "grad_norm": 0.07416783273220062, + "kept_token_count": 573.41875, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.983730149269104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.355425539368298e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.54688339233398, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8543640136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27820, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30014343559741974, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.012218147795647383, + "grad_norm": 0.08605493605136871, + "kept_token_count": 572.65, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9815554678440094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.204914290923626e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.5525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9321350097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27830, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3000726461410522, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.009438313124701381, + "grad_norm": 0.15623845160007477, + "kept_token_count": 572.646875, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9817179203033447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.141455935197883e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.75430755615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5263153076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27840, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.29931963980197906, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.010798076773062349, + "grad_norm": 0.054698169231414795, + "kept_token_count": 573.40625, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9830690741539001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.613662117160857e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.36747894287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.70404052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30060418546199796, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.010007570823654532, + "grad_norm": 0.08356853574514389, + "kept_token_count": 572.4796875, + "learning_rate": 4.158032615452251e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9831153631210328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.595583981834352e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.68758697509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.31061096191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27860, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.30047853887081144, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.01153173097409308, + "grad_norm": 0.11042077094316483, + "kept_token_count": 572.33125, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.982409805059433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.871188670629636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.97842941284179, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.47222595214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27870, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.3002281039953232, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.012034891871735454, + "grad_norm": 0.059552475810050964, + "kept_token_count": 572.540625, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9835767030715943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.41536735201953e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.95015335083008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3473754882813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27880, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3009187191724777, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.012685497850179672, + "grad_norm": 0.13481754064559937, + "kept_token_count": 571.76875, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9832521557807923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.542145529238041e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.35647583007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.57484436035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27890, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.2992011934518814, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.012061411887407303, + "grad_norm": 0.11883607506752014, + "kept_token_count": 573.4703125, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9830645322799683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.615441961912438e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.65859756469726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.24644165039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27900, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30014358162879945, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.012995756138116122, + "grad_norm": 0.102115698158741, + "kept_token_count": 572.5, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9827371537685394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.743316116626374e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.04449844360352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1851043701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27910, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.3010824263095856, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.013100583059713244, + "grad_norm": 0.09819997847080231, + "kept_token_count": 571.74375, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9830559313297271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.618792140216101e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.56608963012695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7531677246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27920, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30056132674217223, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.010287730488926172, + "grad_norm": 0.10505766421556473, + "kept_token_count": 572.3109375, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9825266242027283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.82555644743843e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.25866394042968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.62010803222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27930, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.30052470266819, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.011226270534098148, + "grad_norm": 0.1003977507352829, + "kept_token_count": 572.36875, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9834030747413636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.483193101303186e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.57587051391602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.06410827636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30120514035224916, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.010427864361554384, + "grad_norm": 0.1431569755077362, + "kept_token_count": 571.6828125, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9834751129150391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.455052171077114e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.94355392456055, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3848907470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27950, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30011407732963563, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.0114209801889956, + "grad_norm": 0.12342024594545364, + "kept_token_count": 572.578125, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0115, + "mgd_cos_sim": 0.9832175195217132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.555669315275737e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.8236213684082, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7151672363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27960, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.30032731890678405, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.010603692336007952, + "grad_norm": 0.10047423094511032, + "kept_token_count": 572.5234375, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9821277320384979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.981373298913241e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.16450119018555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1884765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.29955604374408723, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.011268467456102372, + "grad_norm": 0.07704663276672363, + "kept_token_count": 572.971875, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9827458024024963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.739942546118982e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.55919036865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.13469848632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27980, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.2989500403404236, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.012536660023033618, + "grad_norm": 0.0715366005897522, + "kept_token_count": 573.6421875, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9823555409908294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.892383316881023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.98125762939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7099273681641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27990, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2995506823062897, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.012046751007437706, + "grad_norm": 0.0858864113688469, + "kept_token_count": 573.215625, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9827618360519409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.733677982992959e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01710433959961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5285247802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28000, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29944653809070587, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.013329878123477102, + "grad_norm": 0.08191090822219849, + "kept_token_count": 573.240625, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9828649580478668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.693393297609873e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21658630371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5441070556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28010, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.29945509433746337, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.011801915150135756, + "grad_norm": 0.08801287412643433, + "kept_token_count": 573.21875, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.983866834640503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.30203794571571e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.89345016479493, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3542846679687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29959448277950285, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.009515497554093599, + "grad_norm": 0.10479636490345001, + "kept_token_count": 573.1640625, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9826441645622254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.7796422081301e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.0146255493164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.07309875488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28030, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.3001169979572296, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.011978916451334953, + "grad_norm": 0.11081354320049286, + "kept_token_count": 572.6546875, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9840635597705841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.225186589290388e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.14133071899414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.36048583984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28040, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2999887436628342, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.012045155512169003, + "grad_norm": 0.0691160187125206, + "kept_token_count": 572.8484375, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9829601407051086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.656212244706693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91336288452149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7856048583984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28050, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2996577531099319, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.011988674523308873, + "grad_norm": 0.11149103939533234, + "kept_token_count": 573.05625, + "learning_rate": 3.420305900743942e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9834346532821655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.470852094935254e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.64989624023437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.65553894042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28060, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.299398672580719, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.009634006163105368, + "grad_norm": 0.07533348351716995, + "kept_token_count": 573.4015625, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9835300445556641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.433593080146238e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.70182037353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0214447021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28070, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29985884130001067, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.01243909983895719, + "grad_norm": 0.12471114844083786, + "kept_token_count": 572.734375, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9826338648796081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.78366828651633e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.55791931152343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.05442504882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28080, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.3005610227584839, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.012194722285494208, + "grad_norm": 0.1304420530796051, + "kept_token_count": 572.3171875, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9836780250072479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.375789089361206e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10403366088867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3143829345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28090, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.30008420944213865, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.010505949845537543, + "grad_norm": 0.13497938215732574, + "kept_token_count": 572.71875, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9835436701774597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.428268388845027e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.73668060302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.08058776855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28100, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3006930261850357, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.013938757870346308, + "grad_norm": 0.10392682999372482, + "kept_token_count": 572.2078125, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9821015298366547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.991606314841193e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9563606262207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0806030273437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28110, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.30075020492076876, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.010902606975287199, + "grad_norm": 0.08763566613197327, + "kept_token_count": 572.1078125, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.983443284034729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.467487546615302e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.86187362670898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.27207336425784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28120, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29961897134780885, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.011752458522096276, + "grad_norm": 0.06497742235660553, + "kept_token_count": 573.2609375, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9833229899406433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.514477463497315e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.21978225708008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8352020263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28130, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29934712648391726, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.012000004854053259, + "grad_norm": 0.05785754695534706, + "kept_token_count": 573.3953125, + "learning_rate": 3.14514360460183e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9840562164783477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.228061101865023e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.09673843383788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9989929199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30121409595012666, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.01155428341589868, + "grad_norm": 0.10146552324295044, + "kept_token_count": 571.8, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9831661999225616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.575713559868746e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.13157272338867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7140777587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28150, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2996807605028152, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.011860833223909139, + "grad_norm": 0.09555359929800034, + "kept_token_count": 573.146875, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9827178120613098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.750869506504387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.29602127075195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6936065673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28160, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.3000546216964722, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.01178498831577599, + "grad_norm": 0.07668758928775787, + "kept_token_count": 572.7046875, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9827777802944183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.727438594680279e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34612350463867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.01673278808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28170, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30092457234859465, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.01128969369456172, + "grad_norm": 0.14021024107933044, + "kept_token_count": 572.140625, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9830101370811463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.636683174292557e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.37758407592773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8724731445312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2996377021074295, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.011757645336911083, + "grad_norm": 0.09409455955028534, + "kept_token_count": 572.8828125, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9833903610706329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.488160215667449e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.24844207763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0081848144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28190, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.29929983615875244, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.008464806200936436, + "grad_norm": 0.07424185425043106, + "kept_token_count": 573.2890625, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0085, + "mgd_cos_sim": 0.9832431375980377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.545667056343519e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01138153076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.69496459960936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2997090995311737, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.01025056019425392, + "grad_norm": 0.0822618380188942, + "kept_token_count": 573.08125, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9828189492225647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.711365713272244e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.99869766235352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.323876953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28210, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29886201322078704, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.00996573232114315, + "grad_norm": 0.10530417412519455, + "kept_token_count": 574.084375, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.01, + "mgd_cos_sim": 0.983501935005188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.444578393711708e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09883117675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.509033203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28220, + "valid_token_count": 818.796875 + }, + { + "actual_token_mask_ratio": 0.3000542879104614, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.008898727968335152, + "grad_norm": 0.09320589900016785, + "kept_token_count": 572.7859375, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0089, + "mgd_cos_sim": 0.9833446025848389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.506032368633897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.61519393920898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1444091796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28230, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.29906007647514343, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.01160724605433643, + "grad_norm": 0.0898299291729927, + "kept_token_count": 573.65, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9830365717411041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.626359026995488e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.74765167236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3547332763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28240, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2993498921394348, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.011613070219755172, + "grad_norm": 0.08999185264110565, + "kept_token_count": 573.24375, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9833603143692017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.499895243905485e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.12877731323242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.74913330078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28250, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2996825575828552, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.012443687580525876, + "grad_norm": 0.08805599808692932, + "kept_token_count": 572.9640625, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9835059165954589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.443019847210963e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.84754180908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.77691345214845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28260, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29892762899398806, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.01220743702724576, + "grad_norm": 0.13928043842315674, + "kept_token_count": 573.71875, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9830261111259461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.63043960230425e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.49405364990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.59059753417966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28270, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.30009315609931947, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.011542989779263735, + "grad_norm": 0.10033346712589264, + "kept_token_count": 572.4890625, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9829352915287017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.66591469780542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.66384048461914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8031890869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28280, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.30096350610256195, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.011129686189815403, + "grad_norm": 0.07630517333745956, + "kept_token_count": 572.0, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9838701665401459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.300735512922983e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.46068267822265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4435150146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28290, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30029651820659636, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.01114615760743618, + "grad_norm": 0.07519548386335373, + "kept_token_count": 572.471875, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9825445115566254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.818572474003304e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.41349258422852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7230285644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28300, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.300720340013504, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.014063670393079519, + "grad_norm": 0.11276088654994965, + "kept_token_count": 572.3125, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.983138358592987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.5865982833202e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.52453155517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.60381469726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28310, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30048492550849915, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.010297630494460464, + "grad_norm": 0.06723912805318832, + "kept_token_count": 572.109375, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9830998301506042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.601646055059973e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.6324577331543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6349609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28320, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.3002092868089676, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.009989575576037168, + "grad_norm": 0.09588795155286789, + "kept_token_count": 572.54375, + "learning_rate": 2.537440882031683e-07, + "loss": 0.01, + "mgd_cos_sim": 0.982692813873291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.760635151294991e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.43098220825195, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4709777832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28330, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.3001154989004135, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.012053304072469472, + "grad_norm": 0.1424655318260193, + "kept_token_count": 572.7546875, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9820683300495148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.004576618783176e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.2690933227539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6941284179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28340, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2981913685798645, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.010466557368636132, + "grad_norm": 0.07847106456756592, + "kept_token_count": 574.4515625, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9830648005008698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.615338352276013e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.55971450805664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.27118225097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28350, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.2996696323156357, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.011951899901032449, + "grad_norm": 0.07406924664974213, + "kept_token_count": 572.896875, + "learning_rate": 2.447394127561181e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9820569157600403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.009031905909069e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.00459747314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9146240234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28360, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2992981195449829, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.010024250205606221, + "grad_norm": 0.0977407917380333, + "kept_token_count": 573.175, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9821720004081727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.964080494071822e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.42744522094726, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9991485595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28370, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.30057761669158933, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.011404458060860634, + "grad_norm": 0.07020827382802963, + "kept_token_count": 572.1296875, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9829046845436096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.677878336631693e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.51938323974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8902923583984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28380, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.29949283599853516, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.011416208557784557, + "grad_norm": 0.12205301970243454, + "kept_token_count": 573.2015625, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9840858221054077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.216496585693676e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.22209396362305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5419921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28390, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.3010202914476395, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.01013132007792592, + "grad_norm": 0.09233621507883072, + "kept_token_count": 572.0546875, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.982390958070755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.878550557303243e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.81169204711914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.83977661132815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28400, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.299353688955307, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.011093915160745382, + "grad_norm": 0.07741901278495789, + "kept_token_count": 573.396875, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9823191344738007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.906606577103958e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.73473205566407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9437286376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28410, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30001685917377474, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.009742041025310755, + "grad_norm": 0.07117502391338348, + "kept_token_count": 572.96875, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9830920398235321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.604691225220449e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.43334350585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28420, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.30165922045707705, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.010625718673691154, + "grad_norm": 0.07277233898639679, + "kept_token_count": 571.48125, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9832570612430572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.540227695950307e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4265853881836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.64765014648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30044137239456176, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.01007303325459361, + "grad_norm": 0.1275552213191986, + "kept_token_count": 572.553125, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9830416917800904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.624359048146289e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7512596130371, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.26524963378904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28440, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.3005512773990631, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.009859925834462047, + "grad_norm": 0.09758784621953964, + "kept_token_count": 572.465625, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9835400640964508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.429679051507264e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.22271347045898, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8429351806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.29987818002700806, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.009954193606972694, + "grad_norm": 0.08697658032178879, + "kept_token_count": 573.071875, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9825153768062591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.829951707914006e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.54836044311523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.98802490234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28460, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.3003358870744705, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.01154882051050663, + "grad_norm": 0.09362097829580307, + "kept_token_count": 572.3609375, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9825197279453277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.828252917330246e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86208267211914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3103454589844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28470, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.3001276284456253, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.01208400446921587, + "grad_norm": 0.11189528554677963, + "kept_token_count": 572.7359375, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9830653846263886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.615108068217524e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.76427536010742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92108154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29883230328559873, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.009417021926492453, + "grad_norm": 0.08445744216442108, + "kept_token_count": 573.675, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9826991140842438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.758180643373635e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91754837036133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5951293945312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28490, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2993814587593079, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.01354113849811256, + "grad_norm": 0.1054430827498436, + "kept_token_count": 573.390625, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9823070883750915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.911308119015303e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.87602310180664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1813568115234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28500, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.3011725902557373, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.010854433942586183, + "grad_norm": 0.0643489807844162, + "kept_token_count": 571.7109375, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.982585608959198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.802507450629492e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.77127685546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1484680175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28510, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.2994529902935028, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.009436147660017014, + "grad_norm": 0.05894291028380394, + "kept_token_count": 573.4140625, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9833641231060029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.498405782622285e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.81628646850587, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9880645751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28520, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.3000631660223007, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.009712068596854806, + "grad_norm": 0.0723186731338501, + "kept_token_count": 572.675, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.983963668346405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.264211442612578e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9809455871582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8083862304687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28530, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.30050204396247865, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.010827563237398863, + "grad_norm": 0.08455909788608551, + "kept_token_count": 572.4578125, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9835139572620392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.43988372758031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.09733963012695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3129119873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28540, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2994868099689484, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.010256853001192213, + "grad_norm": 0.08526907116174698, + "kept_token_count": 573.0921875, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9823321282863617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.901534652570263e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.75639495849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3135559082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28550, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2997839331626892, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.009905101265758276, + "grad_norm": 0.09814417362213135, + "kept_token_count": 573.1171875, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9838176012039185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.321268738247454e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.05649642944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.04741821289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28560, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.30063320994377135, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.012244867719709873, + "grad_norm": 0.05810865759849548, + "kept_token_count": 572.359375, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9832700967788697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.535142201755662e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.38846130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8032165527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28570, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3002346307039261, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.013933948474004865, + "grad_norm": 0.11965510994195938, + "kept_token_count": 572.56875, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9836712539196014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.378433026839047e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.74387893676757, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6505981445313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28580, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2996764302253723, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.010423291800543666, + "grad_norm": 0.14771263301372528, + "kept_token_count": 573.046875, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.982951033115387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.659768077952322e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.7175422668457, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4527160644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28590, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.29996957182884215, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.012281367182731628, + "grad_norm": 0.08021961152553558, + "kept_token_count": 573.021875, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9827754855155945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.728338630637153e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.19499206542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.40630493164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28600, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.3006676137447357, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.009900550311431288, + "grad_norm": 0.10845930129289627, + "kept_token_count": 572.3765625, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9834004044532776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.484232435468584e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.92306976318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0141357421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.29928632378578185, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.011239825515076518, + "grad_norm": 0.0638079047203064, + "kept_token_count": 573.415625, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9827136814594268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.752492954547052e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.71432876586914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.02765502929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28620, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3011926800012589, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.011690571159124374, + "grad_norm": 0.0816228985786438, + "kept_token_count": 571.8859375, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9833419263362885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.507081379822922e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.73488845825196, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0831268310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28630, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.29971338212490084, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.011709659593179822, + "grad_norm": 0.10221847891807556, + "kept_token_count": 573.0609375, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9819511413574219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.050356653053314e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.34968414306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.10699462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28640, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29944649040699006, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.00963438064791262, + "grad_norm": 0.08315157145261765, + "kept_token_count": 573.096875, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9831707656383515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.573938808287493e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.74022750854492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9797760009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28650, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2998332053422928, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.011194290407001973, + "grad_norm": 0.10171093791723251, + "kept_token_count": 572.65625, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9825878977775574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.801615563745145e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17851181030274, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.43260192871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28660, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.2998673528432846, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.012007127329707146, + "grad_norm": 0.07879777997732162, + "kept_token_count": 572.703125, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9835512518882752, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.425309693440795e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.50369033813476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.67254028320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28670, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.29810814261436464, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.008964299177750944, + "grad_norm": 0.08889663964509964, + "kept_token_count": 574.4703125, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.009, + "mgd_cos_sim": 0.9834244847297668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.474826550402213e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.10201950073242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0385284423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28680, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.30004519522190093, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.011137962201610207, + "grad_norm": 0.09484223276376724, + "kept_token_count": 572.784375, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9831407487392425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.585667542822194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.92633819580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.12778930664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28690, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.30022874772548674, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.011027025198563934, + "grad_norm": 0.1260450929403305, + "kept_token_count": 572.7375, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9829692125320435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.652669726463501e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.92981185913087, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.63402404785154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28700, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.2986218512058258, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.010327033791691064, + "grad_norm": 0.09215296804904938, + "kept_token_count": 573.8375, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9828183352947235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.71159887133399e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.00755081176757, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3449737548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28710, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3004962295293808, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.011953653814271093, + "grad_norm": 0.08875308930873871, + "kept_token_count": 572.3953125, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9824919641017914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.839091365691274e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.80655364990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.17491455078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28720, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.300844219326973, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.009782708622515202, + "grad_norm": 0.061832040548324585, + "kept_token_count": 572.384375, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9829098522663117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.675852127955296e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.36725234985352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1702880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28730, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.3010713040828705, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.010629733605310322, + "grad_norm": 0.0713997334241867, + "kept_token_count": 571.9078125, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9836045861244201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.404480118362698e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.20903778076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8643096923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2997584670782089, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.01207214705646038, + "grad_norm": 0.09856466203927994, + "kept_token_count": 572.875, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9834198415279388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.47664644930046e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.63309631347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.01137390136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28750, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.30013829469680786, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.012365424213930965, + "grad_norm": 0.07298222184181213, + "kept_token_count": 572.5953125, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0124, + "mgd_cos_sim": 0.9830870926380157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.606624810956419e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.6688232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5840698242188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28760, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.298934668302536, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.01046911282464862, + "grad_norm": 0.09457379579544067, + "kept_token_count": 573.88125, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9829194962978363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.672093768429476e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.86192474365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.95542297363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28770, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.2989952743053436, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.012158856680616736, + "grad_norm": 0.0804222822189331, + "kept_token_count": 573.6765625, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9824193894863129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.867441843496635e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.77789535522462, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.37500915527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28780, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30022206008434293, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.010015869326889516, + "grad_norm": 0.13782402873039246, + "kept_token_count": 572.59375, + "learning_rate": 1.334487671266116e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9830281734466553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.629637828154955e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.02479629516601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4988586425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28790, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29980588555335996, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.010836525727063418, + "grad_norm": 0.09363914281129837, + "kept_token_count": 572.8484375, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.983205235004425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.560473011631984e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.16818618774414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1739227294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2995468497276306, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.011910662613809108, + "grad_norm": 0.11861655116081238, + "kept_token_count": 573.0984375, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9827213943004608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.749471504008397e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.0988265991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.07451782226565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28810, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.29901939630508423, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.01306091295555234, + "grad_norm": 0.09747961908578873, + "kept_token_count": 573.66875, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9834872901439666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.450298569689039e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.58496170043945, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0827697753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29964718222618103, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.012297854106873275, + "grad_norm": 0.10094598680734634, + "kept_token_count": 573.0984375, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9825125396251678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.831057216913906e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.01128540039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1228729248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29954774081707003, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.011172517156228422, + "grad_norm": 0.10492794215679169, + "kept_token_count": 573.409375, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9827100276947022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.753915258741471e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.01050567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0098937988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28840, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.2990775555372238, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.011003797221928836, + "grad_norm": 0.07482751458883286, + "kept_token_count": 573.409375, + "learning_rate": 1.205700016610395e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9835028409957886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.444218197430019e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.90128631591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5828796386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28850, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.30095587968826293, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.011535532725974917, + "grad_norm": 0.15328234434127808, + "kept_token_count": 572.034375, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9838398158550262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.312581281235907e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.63238067626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.11549377441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28860, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.29918740689754486, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.014493812248110772, + "grad_norm": 0.12470073997974396, + "kept_token_count": 573.75, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9826800346374511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.765639009245205e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.49878540039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.82171630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28870, + "valid_token_count": 818.6953125 + }, + { + "actual_token_mask_ratio": 0.2992280602455139, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.011716934759169817, + "grad_norm": 0.13840724527835846, + "kept_token_count": 573.6046875, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9826842308044433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.76399275107542e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.35868835449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.06796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28880, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.2999631196260452, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.01022060913965106, + "grad_norm": 0.0944366306066513, + "kept_token_count": 572.7171875, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9829754769802094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.650227951467968e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.97645950317383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.417333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28890, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.3000964641571045, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.012210406037047506, + "grad_norm": 0.07331810146570206, + "kept_token_count": 572.6546875, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9824195444583893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.867386910016648e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.30707015991212, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.60814514160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28900, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30060833096504214, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.013026230968534946, + "grad_norm": 0.09930198639631271, + "kept_token_count": 572.38125, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9832789123058319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.53168997814646e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.72770767211914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4171112060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29931193590164185, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.010524302301928402, + "grad_norm": 0.10127965360879898, + "kept_token_count": 573.396875, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9839166939258576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.282561516854913e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.71032257080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9295928955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28920, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.299140653014183, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.011743825580924749, + "grad_norm": 0.19407087564468384, + "kept_token_count": 573.4515625, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9829964339733124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.642033949901815e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17835922241211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.85169372558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28930, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3006830781698227, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.010716828610748052, + "grad_norm": 0.08856375515460968, + "kept_token_count": 572.25625, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9831696927547455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.574358121724799e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.8214828491211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9905120849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28940, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.3000513046979904, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.013779027294367552, + "grad_norm": 0.055857688188552856, + "kept_token_count": 572.76875, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9831076383590698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.598599429707974e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.69869384765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6005493164063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28950, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3000528275966644, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.012594258785247803, + "grad_norm": 0.07973825186491013, + "kept_token_count": 572.5390625, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0126, + "mgd_cos_sim": 0.9834399223327637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.468797582783736e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.94807586669921, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5482971191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28960, + "valid_token_count": 817.971875 + }, + { + "actual_token_mask_ratio": 0.2992636948823929, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.010738555109128356, + "grad_norm": 0.06634029000997543, + "kept_token_count": 573.4546875, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9823179483413697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.907077549840323e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.11800994873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2798583984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28970, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2994658023118973, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.0116573054343462, + "grad_norm": 0.1306888610124588, + "kept_token_count": 573.08125, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0117, + "mgd_cos_sim": 0.9829298615455627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.668041278317105e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.90094375610352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6669586181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28980, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.300573006272316, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.010301392013207077, + "grad_norm": 0.07902975380420685, + "kept_token_count": 572.0640625, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0103, + "mgd_cos_sim": 0.9831414341926574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.585397823073435e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9965591430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.42564697265624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28990, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.29901168048381804, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.010151993995532394, + "grad_norm": 0.07737360894680023, + "kept_token_count": 573.4078125, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0102, + "mgd_cos_sim": 0.983703225851059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.365942390402779e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.14532623291015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0426818847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29000, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.2992475688457489, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.011286359559744597, + "grad_norm": 0.09301003068685532, + "kept_token_count": 573.19375, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.9830369710922241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.626203648920636e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.96996688842773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5115234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29010, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2991849422454834, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.012172886356711387, + "grad_norm": 0.1092570573091507, + "kept_token_count": 573.6484375, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0122, + "mgd_cos_sim": 0.9828929841518402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.682440798613242e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7947868347168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.61419067382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29020, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.30082679986953736, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.011093228729441762, + "grad_norm": 0.09273520112037659, + "kept_token_count": 571.7734375, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9824230432510376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.866017829452176e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.27061309814454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5223327636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29030, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.29996390640735626, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.01134961498901248, + "grad_norm": 0.10196873545646667, + "kept_token_count": 572.6734375, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0114, + "mgd_cos_sim": 0.9831273436546326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.59089746477548e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.95304336547852, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.15350646972655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29040, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.2993855893611908, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.012535807862877845, + "grad_norm": 0.06538617610931396, + "kept_token_count": 573.4671875, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0126, + "mgd_cos_sim": 0.9823994576931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.875235230836551e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.73167724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9157196044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29050, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.29995833337306976, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.012219715071842075, + "grad_norm": 0.08680766075849533, + "kept_token_count": 572.6140625, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0123, + "mgd_cos_sim": 0.9829402565956116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.663983294856735e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.91978302001954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29060, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.2998128682374954, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.011989046167582274, + "grad_norm": 0.11966587603092194, + "kept_token_count": 572.89375, + "learning_rate": 7.89204318007003e-08, + "loss": 0.012, + "mgd_cos_sim": 0.9817547976970673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.127058488549664e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.48914642333985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.29266052246095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29070, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2992654800415039, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.010933853825554252, + "grad_norm": 0.09176645427942276, + "kept_token_count": 573.559375, + "learning_rate": 7.723559506062849e-08, + "loss": 0.011, + "mgd_cos_sim": 0.9821097314357757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.988395180087536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.62392196655273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3050231933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29080, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.30007848143577576, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.009843225684016943, + "grad_norm": 0.0831635594367981, + "kept_token_count": 572.6703125, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0099, + "mgd_cos_sim": 0.9830502271652222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.621027678193059e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21586761474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7026885986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29090, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30054462552070615, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.01110350047238171, + "grad_norm": 0.06029998138546944, + "kept_token_count": 572.1140625, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9832693219184876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.535438587889076e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.67948913574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6780731201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29100, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.3000067800283432, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.011631051683798433, + "grad_norm": 0.10398121923208237, + "kept_token_count": 572.9984375, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0117, + "mgd_cos_sim": 0.9830166280269623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.634149140154477e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.76638412475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0130310058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29110, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.29926871359348295, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.014054097095504403, + "grad_norm": 0.12724906206130981, + "kept_token_count": 573.3109375, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9824654400348664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.849460332887247e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.07206268310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9414581298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29120, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3003453463315964, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.01222055871039629, + "grad_norm": 0.08100477606058121, + "kept_token_count": 572.4828125, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0123, + "mgd_cos_sim": 0.9817964673042298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.110773294698447e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.45843353271485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.34808959960935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29130, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.3002412647008896, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.01232057656161487, + "grad_norm": 0.07758281379938126, + "kept_token_count": 572.5859375, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0124, + "mgd_cos_sim": 0.9834813416004181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.452614397858269e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.47837982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.03540954589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29140, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2999560534954071, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.011600222159177064, + "grad_norm": 0.06937875598669052, + "kept_token_count": 572.88125, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0116, + "mgd_cos_sim": 0.9835695922374725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.418147422664333e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.14966049194337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.96363830566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29150, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30004655122756957, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.011177702946588397, + "grad_norm": 0.07943335175514221, + "kept_token_count": 572.590625, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.9831321239471436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.589026124856901e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.79624710083007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3553985595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29160, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.2997395217418671, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.0112212176900357, + "grad_norm": 0.0668550506234169, + "kept_token_count": 572.9703125, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.983295351266861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.525274366140366e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.43061447143555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.13902587890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29170, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3001139163970947, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.012831023288890719, + "grad_norm": 0.10236163437366486, + "kept_token_count": 572.728125, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0129, + "mgd_cos_sim": 0.9828562080860138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.696816417388618e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.23121719360351, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3561614990234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29180, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.29912897646427156, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.010644826805219054, + "grad_norm": 0.0807325541973114, + "kept_token_count": 573.5921875, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0107, + "mgd_cos_sim": 0.9820731461048127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.002694001130294e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.75075912475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7978210449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29190, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2996072471141815, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.009096397319808602, + "grad_norm": 0.08326536417007446, + "kept_token_count": 573.0140625, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0091, + "mgd_cos_sim": 0.983282458782196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.530305981868877e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.26560516357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9045074462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29200, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.3007864147424698, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.011159340338781475, + "grad_norm": 0.14358723163604736, + "kept_token_count": 572.2765625, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.9836299598217011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.394567171810194e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.60467071533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3988525390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29210, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.2999193757772446, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.011538480408489704, + "grad_norm": 0.10489063709974289, + "kept_token_count": 572.8703125, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0116, + "mgd_cos_sim": 0.9836122512817382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.401481696229893e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.4413558959961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.52825622558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29220, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3010912984609604, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.011074813595041633, + "grad_norm": 0.08922070264816284, + "kept_token_count": 571.853125, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9825885534286499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.801367271691561e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.1109619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.24297790527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30023738741874695, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.011388740222901105, + "grad_norm": 0.04979180544614792, + "kept_token_count": 573.053125, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0114, + "mgd_cos_sim": 0.9827705323696136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.730279128532856e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.7242202758789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0410888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29240, + "valid_token_count": 818.9203125 + }, + { + "actual_token_mask_ratio": 0.3004559248685837, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.010557501250877977, + "grad_norm": 0.08343259990215302, + "kept_token_count": 572.4171875, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0106, + "mgd_cos_sim": 0.9834754586219787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.454923350247554e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.17047119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16376342773435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29250, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30014423429965975, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.012021881109103561, + "grad_norm": 0.09003312140703201, + "kept_token_count": 572.703125, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0121, + "mgd_cos_sim": 0.9834616243839264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.46032342046965e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.2543960571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.53078002929686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29260, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.29960829615592954, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.012105459114536643, + "grad_norm": 0.1172274574637413, + "kept_token_count": 573.0671875, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0121, + "mgd_cos_sim": 0.9829117178916931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.67513049847912e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.71489791870117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7784118652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29270, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29975863099098204, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.011714680446311831, + "grad_norm": 0.12111372500658035, + "kept_token_count": 573.0421875, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0117, + "mgd_cos_sim": 0.9830283284187317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.629577037529089e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.82590942382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4178894042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29280, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29977398812770845, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.012037587631493808, + "grad_norm": 0.10049507766962051, + "kept_token_count": 572.74375, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0121, + "mgd_cos_sim": 0.9824998915195465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.836000393377617e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.959326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9107147216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29290, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.30000867545604704, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.010549278720282019, + "grad_norm": 0.13111017644405365, + "kept_token_count": 572.6703125, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0106, + "mgd_cos_sim": 0.9824775218963623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.844739109510556e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.44110565185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8025726318359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29300, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29974593222141266, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.010931139485910535, + "grad_norm": 0.08820366114377975, + "kept_token_count": 573.109375, + "learning_rate": 4.349280298500291e-08, + "loss": 0.011, + "mgd_cos_sim": 0.9813395738601685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.289250788744539e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49800109863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2396667480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29310, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.300368469953537, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.013204717403277754, + "grad_norm": 0.10952293127775192, + "kept_token_count": 572.4765625, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0132, + "mgd_cos_sim": 0.9828285098075866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.707632201141678e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.92186737060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5895233154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29320, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29995799958705904, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.010085816029459238, + "grad_norm": 0.0763789713382721, + "kept_token_count": 572.803125, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0101, + "mgd_cos_sim": 0.9823996186256408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.87516301695723e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.63758544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9063262939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29330, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3006862193346024, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.012461097911000252, + "grad_norm": 0.09221675246953964, + "kept_token_count": 572.3875, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0125, + "mgd_cos_sim": 0.9832009553909302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.562144080817234e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.85543746948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5220031738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2997701644897461, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.012597685027867556, + "grad_norm": 0.07395664602518082, + "kept_token_count": 573.015625, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0126, + "mgd_cos_sim": 0.9831110179424286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.597274659725372e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.46558837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3284881591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29977332055568695, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.01116815865971148, + "grad_norm": 0.07501410692930222, + "kept_token_count": 572.8953125, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.982051146030426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.011286870692857e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.83189163208007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8723602294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29360, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3000007659196854, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.011098805535584687, + "grad_norm": 0.1019258126616478, + "kept_token_count": 573.0515625, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9823846638202667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.881004155729897e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.44795837402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7037017822266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29370, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.29954497814178466, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.010425922507420183, + "grad_norm": 0.09964429587125778, + "kept_token_count": 573.271875, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0105, + "mgd_cos_sim": 0.9824286699295044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.86382052663248e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90014343261718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8980712890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29380, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3000373005867004, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.010856215702369809, + "grad_norm": 0.09114211052656174, + "kept_token_count": 572.628125, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0109, + "mgd_cos_sim": 0.9830043375492096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.638946251769084e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.53400650024415, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0684295654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29390, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.29924190044403076, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.01104733431711793, + "grad_norm": 0.11131253838539124, + "kept_token_count": 573.3640625, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9832835555076599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.529880847665482e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.36692428588867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.93470458984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29400, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29994343817234037, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.010746206808835267, + "grad_norm": 0.06410326808691025, + "kept_token_count": 572.871875, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9835952460765839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.408126901078504e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.24301223754883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.76568603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29410, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.30033365786075594, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.011267748661339283, + "grad_norm": 0.17413631081581116, + "kept_token_count": 572.4078125, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.9832911014556884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.526932193082758e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.3277976989746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7975555419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29420, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.30127093493938445, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.014736432302743196, + "grad_norm": 0.13208429515361786, + "kept_token_count": 571.9984375, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0148, + "mgd_cos_sim": 0.9826266050338746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.78649616020266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.12556381225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.88750915527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29430, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.29907116293907166, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.008204739168286323, + "grad_norm": 0.0632253959774971, + "kept_token_count": 573.3859375, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0082, + "mgd_cos_sim": 0.9830679953098297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614082267333287e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.85331878662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1718017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29440, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30023076236248014, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.010124443238601089, + "grad_norm": 0.0826222375035286, + "kept_token_count": 572.609375, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0102, + "mgd_cos_sim": 0.983549153804779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.426129839383065e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.67505569458008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.29807434082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29450, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.29953135550022125, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.013927343115210534, + "grad_norm": 0.1647743284702301, + "kept_token_count": 573.246875, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.014, + "mgd_cos_sim": 0.982217800617218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.946194371266756e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.86570510864257, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.11485290527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29460, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.3002903193235397, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.01089539285749197, + "grad_norm": 0.07641911506652832, + "kept_token_count": 572.5796875, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0109, + "mgd_cos_sim": 0.98275306224823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.737107723893132e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.96548614501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.51393432617186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29470, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3001138985157013, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.01114072329364717, + "grad_norm": 0.09972655773162842, + "kept_token_count": 572.6125, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.9821868062019348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.958300255064387e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.58550033569335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3147308349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29480, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2999516248703003, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.012359747756272554, + "grad_norm": 0.09081368893384933, + "kept_token_count": 572.865625, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0124, + "mgd_cos_sim": 0.9829185009002686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.672481831628829e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96042938232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7907012939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29490, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.2993578642606735, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.011858284147456289, + "grad_norm": 0.10562199354171753, + "kept_token_count": 573.1765625, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0119, + "mgd_cos_sim": 0.9823556125164032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.892355377203785e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.43941040039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3464752197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29500, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29909464716911316, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.010933563159778715, + "grad_norm": 0.11911959201097488, + "kept_token_count": 573.6875, + "learning_rate": 2.196485058207298e-08, + "loss": 0.011, + "mgd_cos_sim": 0.983297199010849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.52454651572043e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.31453018188476, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.72173767089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29510, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.300587472319603, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.011353161185979843, + "grad_norm": 0.07901731133460999, + "kept_token_count": 572.0453125, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0114, + "mgd_cos_sim": 0.9821283280849457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.981145852478222e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.47355728149414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.25449523925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29520, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.29961597323417666, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.010692410916090012, + "grad_norm": 0.048407673835754395, + "kept_token_count": 572.978125, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0107, + "mgd_cos_sim": 0.9835041463375092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.443715428758879e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.6244400024414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.958251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29530, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.29977092146873474, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.011702730879187584, + "grad_norm": 0.0779406875371933, + "kept_token_count": 572.9171875, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0117, + "mgd_cos_sim": 0.982765656709671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.732182810083031e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.62063446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3820526123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29540, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29986721873283384, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.010087702982127666, + "grad_norm": 0.09667262434959412, + "kept_token_count": 572.9765625, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0101, + "mgd_cos_sim": 0.9822898387908936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.918046674400102e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.98382568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8420349121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29550, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2999581605195999, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.011384366173297166, + "grad_norm": 0.09541652351617813, + "kept_token_count": 572.85625, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0114, + "mgd_cos_sim": 0.9825655817985535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.810337217757478e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.43884048461913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.50917663574216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29560, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3001148164272308, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.010195839311927557, + "grad_norm": 0.09345252066850662, + "kept_token_count": 572.646875, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0102, + "mgd_cos_sim": 0.9834533929824829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.463532845373266e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.14207382202149, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.24305114746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29570, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2996662139892578, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.013645878899842501, + "grad_norm": 0.0982549637556076, + "kept_token_count": 573.15625, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0137, + "mgd_cos_sim": 0.9830890536308289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.605857706745155e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.34484405517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8140014648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29580, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2988386660814285, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.01075568743981421, + "grad_norm": 0.12888389825820923, + "kept_token_count": 573.809375, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9830656886100769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.614985468331725e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.11488723754883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.80707092285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29590, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.29919514060020447, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.011074085952714086, + "grad_norm": 0.08752553910017014, + "kept_token_count": 573.31875, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9825417101383209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.819663685746491e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.45365371704102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7010498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29600, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.30088419616222384, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.01059391014277935, + "grad_norm": 0.10955969244241714, + "kept_token_count": 572.15625, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0106, + "mgd_cos_sim": 0.9830124497413635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.635777317569591e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.40542831420899, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.10955810546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29610, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.30174965858459474, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.01072042011655867, + "grad_norm": 0.07366969436407089, + "kept_token_count": 571.3578125, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9820371448993683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.01676031894749e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.70473861694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.96381225585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29620, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3001731067895889, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.012208996899425983, + "grad_norm": 0.08738918602466583, + "kept_token_count": 572.603125, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0122, + "mgd_cos_sim": 0.9833395421504975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.508011065307074e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.0816535949707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.61548767089846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29630, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2999570190906525, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.012303641019389033, + "grad_norm": 0.0923442617058754, + "kept_token_count": 572.734375, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0123, + "mgd_cos_sim": 0.9833981692790985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.485109734057914e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.92726974487304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0445281982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29640, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.30060009062290194, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.01161186876706779, + "grad_norm": 0.0998484194278717, + "kept_token_count": 572.4046875, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0116, + "mgd_cos_sim": 0.9831583678722382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.578778593393508e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.49926986694337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.77735290527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29650, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.30021297335624697, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.010551393032073975, + "grad_norm": 0.08991192281246185, + "kept_token_count": 572.7203125, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0106, + "mgd_cos_sim": 0.983799934387207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.328174749796745e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.83558654785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.90704956054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29660, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.30042300522327425, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.009244319144636392, + "grad_norm": 0.10972252488136292, + "kept_token_count": 572.528125, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0093, + "mgd_cos_sim": 0.9837331593036651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.354248980642297e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.87753067016601, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47491760253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29670, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3003954827785492, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.011395400064066053, + "grad_norm": 0.09935993701219559, + "kept_token_count": 572.4765625, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0114, + "mgd_cos_sim": 0.982485693693161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.841549075033981e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.90810623168946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.15308532714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29680, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30079306960105895, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.012297574430704117, + "grad_norm": 0.13097770512104034, + "kept_token_count": 572.1296875, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0123, + "mgd_cos_sim": 0.9834574043750763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.461971206590533e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.71768417358399, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.67796630859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.300566965341568, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.01243923748843372, + "grad_norm": 0.1241481676697731, + "kept_token_count": 572.40625, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0125, + "mgd_cos_sim": 0.9829666078090668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.653690070379525e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.62920227050782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.44309387207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29700, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3004810452461243, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.010721723176538944, + "grad_norm": 0.08025330305099487, + "kept_token_count": 572.3609375, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0108, + "mgd_cos_sim": 0.9828034281730652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.717427459079772e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.24296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5017517089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2999103397130966, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.010393365006893873, + "grad_norm": 0.07378704845905304, + "kept_token_count": 572.8796875, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0104, + "mgd_cos_sim": 0.9837284624576569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.356079684337601e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.72700729370118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2210998535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29720, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29955809712409975, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.010880187060683966, + "grad_norm": 0.09194578975439072, + "kept_token_count": 573.171875, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0109, + "mgd_cos_sim": 0.9832085549831391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.559177490998991e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.59184036254882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.00669860839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30033054351806643, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.009593993984162807, + "grad_norm": 0.10185885429382324, + "kept_token_count": 572.7046875, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0096, + "mgd_cos_sim": 0.9834099471569061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.480502524937037e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.96422805786133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.11253662109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29740, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.30022179484367373, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.011941505875438452, + "grad_norm": 0.11064653843641281, + "kept_token_count": 572.7890625, + "learning_rate": 5.741045996567329e-09, + "loss": 0.012, + "mgd_cos_sim": 0.9826815724372864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.765037323930301e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.26485061645508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29750, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29886743426322937, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.011606285534799098, + "grad_norm": 0.08705411106348038, + "kept_token_count": 573.7890625, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0116, + "mgd_cos_sim": 0.9835365951061249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.431039255403448e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.07979965209961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7050720214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29760, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.30018865168094633, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.011593414796516299, + "grad_norm": 0.09805872291326523, + "kept_token_count": 572.6140625, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0116, + "mgd_cos_sim": 0.9827990770339966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.71912497637095e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.11140365600586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.41717224121095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2988845109939575, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.012572400458157062, + "grad_norm": 0.09351973980665207, + "kept_token_count": 573.7203125, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0126, + "mgd_cos_sim": 0.9825842499732971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.80304703564616e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.3741325378418, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3584716796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29780, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2995338052511215, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.010880154371261597, + "grad_norm": 0.06192271411418915, + "kept_token_count": 573.0984375, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0109, + "mgd_cos_sim": 0.9832721889019013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.534316235047299e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.4084716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8188446044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29790, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.3007198214530945, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.0110284220892936, + "grad_norm": 0.09349194169044495, + "kept_token_count": 571.990625, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0111, + "mgd_cos_sim": 0.9829368829727173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.665296532446519e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.15553970336914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.35883178710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29800, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.30076143741607664, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.011897318018600344, + "grad_norm": 0.09552782028913498, + "kept_token_count": 572.15, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0119, + "mgd_cos_sim": 0.9838236927986145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.318889936665072e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.51249694824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1671508789062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29810, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.29892667531967165, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.011106404662132262, + "grad_norm": 0.09997233748435974, + "kept_token_count": 573.5390625, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0111, + "mgd_cos_sim": 0.9828642308712006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.693681425531394e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.42776260375976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.77101440429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29820, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.30005780756473543, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.01156615037471056, + "grad_norm": 0.10511162877082825, + "kept_token_count": 572.80625, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0116, + "mgd_cos_sim": 0.9832250058650971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.552750382979867e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.31485290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4403961181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29830, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2995928883552551, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.010608728835359215, + "grad_norm": 0.06317409127950668, + "kept_token_count": 573.05625, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0106, + "mgd_cos_sim": 0.9830441296100616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.623399167438038e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.29760513305663, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.00559387207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29840, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2987259656190872, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.01211371785029769, + "grad_norm": 0.08515326678752899, + "kept_token_count": 573.7453125, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0121, + "mgd_cos_sim": 0.983391410112381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.487749524239917e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.64067001342774, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.50482482910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29850, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29959407448768616, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.012670249305665493, + "grad_norm": 0.07106933742761612, + "kept_token_count": 573.3046875, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0127, + "mgd_cos_sim": 0.9835164785385132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.438894160964992e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.18855056762695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9136566162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29860, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.30045621991157534, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.010936436895281076, + "grad_norm": 0.08027641475200653, + "kept_token_count": 572.3, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.011, + "mgd_cos_sim": 0.9836447477340698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.388788387994282e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.49825592041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.59303588867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29870, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.3007035553455353, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.01151617425493896, + "grad_norm": 0.09780815988779068, + "kept_token_count": 572.1296875, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0115, + "mgd_cos_sim": 0.983683830499649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.373521355271806e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.51861343383788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0644165039063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29880, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.3000097185373306, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.012464865203946828, + "grad_norm": 0.08717656880617142, + "kept_token_count": 572.73125, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0125, + "mgd_cos_sim": 0.9841711580753326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.183158038766123e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.84308471679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6978515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29890, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.29994636178016665, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.011409893818199634, + "grad_norm": 0.06425781548023224, + "kept_token_count": 572.9890625, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0114, + "mgd_cos_sim": 0.9818903923034668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 7.074081368045882e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 85.24836578369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8356140136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29900, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.3007087528705597, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.01045452649705112, + "grad_norm": 0.07235594838857651, + "kept_token_count": 572.3546875, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0105, + "mgd_cos_sim": 0.9831718981266022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.573496939381585e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.51385650634765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5632263183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29910, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.300351881980896, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.01153387213125825, + "grad_norm": 0.08397724479436874, + "kept_token_count": 572.403125, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0116, + "mgd_cos_sim": 0.9832250595092773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.552733611897565e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47035675048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7169677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29920, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2996899843215942, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.01460949839092791, + "grad_norm": 0.18218757212162018, + "kept_token_count": 572.9578125, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0146, + "mgd_cos_sim": 0.9821414768695831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.976001313887536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.44074325561523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3952697753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29930, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.29986618757247924, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.010936996201053262, + "grad_norm": 0.08589515835046768, + "kept_token_count": 572.9109375, + "learning_rate": 3.391011967207591e-10, + "loss": 0.011, + "mgd_cos_sim": 0.9830418527126312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.6242952743778e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.22802200317383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6093505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29940, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3010975897312164, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.012068685237318277, + "grad_norm": 0.1498708724975586, + "kept_token_count": 571.890625, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0121, + "mgd_cos_sim": 0.9825876295566559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.801724157412536e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.47996978759765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5728485107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3002112865447998, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.0106069331523031, + "grad_norm": 0.0912938043475151, + "kept_token_count": 572.59375, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0106, + "mgd_cos_sim": 0.9835327804088593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.43253224552609e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 88.29602584838867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7449432373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29960, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.29944680631160736, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.010973935667425393, + "grad_norm": 0.09478842467069626, + "kept_token_count": 573.090625, + "learning_rate": 8.757784487611886e-11, + "loss": 0.011, + "mgd_cos_sim": 0.9831798911094666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.570372752321419e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.93133544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3219940185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29970, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.2990915089845657, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.010129261715337633, + "grad_norm": 0.1880965232849121, + "kept_token_count": 573.340625, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0102, + "mgd_cos_sim": 0.9829979538917542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.641442305408419e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.03892288208007, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0355590820312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29980, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.300194051861763, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.00988436145707965, + "grad_norm": 0.11164721846580505, + "kept_token_count": 572.75, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0099, + "mgd_cos_sim": 0.9837445914745331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.349780960590579e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.39065170288086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.51061096191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29990, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30044817328453066, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.009573338646441698, + "grad_norm": 0.08964826166629791, + "kept_token_count": 572.6296875, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0096, + "mgd_cos_sim": 0.9829586863517761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 6.656781879428308e-05, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.83136672973633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1584716796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30000, + "valid_token_count": 818.5625 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.015038375470290582, + "train_runtime": 57079.6952, + "train_samples_per_second": 67.274, + "train_steps_per_second": 0.526 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..ea413c810f293c53e9485f437e0b8de4225b0051 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54fd331f315b0079b8e5f938a05b82bc6faddf00045a45c9af2fcf22891d59c +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/resolved_config.yaml b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e3d6d412a478ce326d9e9ef846430c933c1f719a --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse/mgd_token_mask_ratio_0.3/resolved_config.yaml @@ -0,0 +1,49 @@ +experiment_name: mgd_v2_loss_mse +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-60000 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only_v2/MGD_v2/mgd_v2_loss_mse +sweep: + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 + seed: 42 +model: null +phases: +- name: phase2_mgd_only + max_steps: 30000 + save_steps: 0 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false + losses: + mgd_enabled: true + mgd_fm_loss_weight: 0.0 + mgd_loss_weight: 1.0 + mgd_sequence_hidden_dim: 512 + mgd_token_mask_ratio: 0.3 + mgd_use_cosine_loss: false + mgd_use_mse_loss: true +- name: phase3_fm_mgd_fixed_0p5 + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: true + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.5 + mgd_token_mask_ratio: 0.3 + mgd_use_cosine_loss: false + mgd_use_mse_loss: true +resolved_sweep: + mgd_token_mask_ratio: 0.3 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c9509993e12e3bc3b291d1a0398318e256c8a2b7 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..713d4351497886cf25fc21e4492db719c3023c92 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:195c04d96fc235692877602d231c57bb35500c91600df74936e5daf4e5adbe63 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..80e273e2655808cd5c400c90125c823acba02a68 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase2/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd4194b69281f2d668f850a7ed0f001b486597afbcae8d89aec953c507e540a2 +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..17a4d2840e3255fe2ad42bb5ebfd613fcb643ac0 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:297ba5d932d6ec1bc6dd6ecc599015a406b97b1ebcb4a815397e60c8d48fa08d +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5485a402c16e61cd6e23161ffab26de5bc7c655b --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2d70c3f40e7be4d09538bde42ca534e9e1c222c358f9189eb025649ee07ebbf +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/trainer_state.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..8757a700a939121f335ac99b910e21e63c2b83b1 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.0995713472366333, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.12954945713281632, + "grad_norm": 0.9835854768753052, + "kept_token_count": 737.025, + "learning_rate": 1.8e-07, + "loss": 0.1334, + "mgd_cos_sim": 0.9922409296035767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007759074307978153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4737152099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.55369873046874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.10046873316168785, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.11920740455389023, + "grad_norm": 0.8481401205062866, + "kept_token_count": 736.1125, + "learning_rate": 3.8e-07, + "loss": 0.1227, + "mgd_cos_sim": 0.9930589377880097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941048800945282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5723449707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.43943481445314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.1001900315284729, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.11050482615828514, + "grad_norm": 0.9211321473121643, + "kept_token_count": 736.375, + "learning_rate": 5.800000000000001e-07, + "loss": 0.114, + "mgd_cos_sim": 0.9930164337158203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006983586307615042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.6455535888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.86651000976565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 30, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.09988440796732903, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.08957599811255931, + "grad_norm": 0.5441311597824097, + "kept_token_count": 736.7640625, + "learning_rate": 7.799999999999999e-07, + "loss": 0.093, + "mgd_cos_sim": 0.9931101024150848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889896001666784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1059844970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3740692138672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 40, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.0996365949511528, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.07996548861265182, + "grad_norm": 0.47049835324287415, + "kept_token_count": 736.728125, + "learning_rate": 9.8e-07, + "loss": 0.0833, + "mgd_cos_sim": 0.9932586133480072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006741380691528321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.1339508056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6266265869141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 50, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.10041718780994416, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.0676284845918417, + "grad_norm": 0.4039013087749481, + "kept_token_count": 735.9796875, + "learning_rate": 1.18e-06, + "loss": 0.0714, + "mgd_cos_sim": 0.9924398839473725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007560115493834019, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.52576599121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5510284423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 60, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.1002819575369358, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.061810869723558426, + "grad_norm": 0.4894472658634186, + "kept_token_count": 736.3359375, + "learning_rate": 1.38e-06, + "loss": 0.0654, + "mgd_cos_sim": 0.9927531599998474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007246837858110667, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.6491943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.64752807617185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 70, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10018862038850784, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.04652238748967648, + "grad_norm": 0.3424137830734253, + "kept_token_count": 736.309375, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.05, + "mgd_cos_sim": 0.9930221796035766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069778251461684706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.02532653808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3195068359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 80, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09996025860309601, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.04559793919324875, + "grad_norm": 0.3494178056716919, + "kept_token_count": 736.334375, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0497, + "mgd_cos_sim": 0.9918069005012512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008193096518516541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.53476867675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.87347717285155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 90, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.10043860003352165, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.038067267276346684, + "grad_norm": 0.2174566239118576, + "kept_token_count": 736.178125, + "learning_rate": 1.98e-06, + "loss": 0.0419, + "mgd_cos_sim": 0.992343294620514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007656715624034405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.384814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7071533203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 100, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.10010955333709717, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.03919001910835505, + "grad_norm": 0.23816710710525513, + "kept_token_count": 736.1921875, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0431, + "mgd_cos_sim": 0.9921159446239471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007884066179394722, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.41490783691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.72942810058595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 110, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.0999799333512783, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.03274309504777193, + "grad_norm": 0.21420703828334808, + "kept_token_count": 736.36875, + "learning_rate": 2.38e-06, + "loss": 0.0369, + "mgd_cos_sim": 0.9917447149753571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00825526863336563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.53572845458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.59771118164065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 120, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09961304664611817, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.028162319213151932, + "grad_norm": 0.16250264644622803, + "kept_token_count": 736.4234375, + "learning_rate": 2.58e-06, + "loss": 0.0322, + "mgd_cos_sim": 0.9919288873672485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008071108348667621, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.9503372192383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9696899414063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 130, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.0997048519551754, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.02995378281921148, + "grad_norm": 0.1760963499546051, + "kept_token_count": 736.7515625, + "learning_rate": 2.78e-06, + "loss": 0.0341, + "mgd_cos_sim": 0.9916913509368896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008308659866452217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.8048583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7611389160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.10025215148925781, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.030330522172152995, + "grad_norm": 0.18096651136875153, + "kept_token_count": 736.165625, + "learning_rate": 2.98e-06, + "loss": 0.0345, + "mgd_cos_sim": 0.991750854253769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008249143417924643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.52326202392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.47815856933596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 150, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.09997637346386909, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.025956650264561176, + "grad_norm": 0.21437548100948334, + "kept_token_count": 736.5265625, + "learning_rate": 3.18e-06, + "loss": 0.0305, + "mgd_cos_sim": 0.9908144652843476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009185525588691235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.7653839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7064544677734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 160, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.10050362348556519, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.023468996584415435, + "grad_norm": 0.18256208300590515, + "kept_token_count": 735.7703125, + "learning_rate": 3.38e-06, + "loss": 0.0281, + "mgd_cos_sim": 0.990665715932846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009334279969334603, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.87862701416014, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0551788330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.10055600479245186, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.020558321848511696, + "grad_norm": 0.1656845062971115, + "kept_token_count": 736.1328125, + "learning_rate": 3.58e-06, + "loss": 0.025, + "mgd_cos_sim": 0.9911175489425659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008882442023605108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.15088958740233, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7500701904297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 180, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.10038567706942558, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.021432589925825595, + "grad_norm": 0.14674465358257294, + "kept_token_count": 736.3046875, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0259, + "mgd_cos_sim": 0.9910051047801971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008994894102215767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.84125518798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4698516845703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 190, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.10023127719759942, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.021748225577175616, + "grad_norm": 0.1753394603729248, + "kept_token_count": 736.1921875, + "learning_rate": 3.98e-06, + "loss": 0.0264, + "mgd_cos_sim": 0.9906255304813385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009374462068080902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.17517852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.88238525390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 200, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.09916112795472146, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.02188669638708234, + "grad_norm": 0.1437690556049347, + "kept_token_count": 737.434375, + "learning_rate": 4.18e-06, + "loss": 0.0271, + "mgd_cos_sim": 0.9896340489387512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010365964379161596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.79598846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.775, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 210, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.1000545397400856, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.02005136329680681, + "grad_norm": 0.12051764875650406, + "kept_token_count": 736.0125, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0248, + "mgd_cos_sim": 0.9904186964035034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009581314492970705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.00595245361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.19049377441405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 220, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.09944932460784912, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.02266224902123213, + "grad_norm": 0.17707450687885284, + "kept_token_count": 736.575, + "learning_rate": 4.58e-06, + "loss": 0.0276, + "mgd_cos_sim": 0.9900985300540924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009901468455791474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.53618774414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6905853271484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 230, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.09986280873417855, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.02468911511823535, + "grad_norm": 0.13770440220832825, + "kept_token_count": 736.28125, + "learning_rate": 4.78e-06, + "loss": 0.0294, + "mgd_cos_sim": 0.9904820561408997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009517954848706722, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.24657440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.63140258789065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 240, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.09930734857916831, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.0198167160153389, + "grad_norm": 0.14063765108585358, + "kept_token_count": 737.2671875, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0247, + "mgd_cos_sim": 0.9902227580547333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009777261968702078, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.14169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9945495605469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 250, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.09957829341292382, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.0190894634462893, + "grad_norm": 0.14166554808616638, + "kept_token_count": 736.8109375, + "learning_rate": 5.18e-06, + "loss": 0.0236, + "mgd_cos_sim": 0.990961080789566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00903891185298562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.52339935302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9767639160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 260, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.10027998462319374, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.02045600926503539, + "grad_norm": 0.12237945944070816, + "kept_token_count": 736.0734375, + "learning_rate": 5.38e-06, + "loss": 0.0253, + "mgd_cos_sim": 0.990274965763092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009725029859691859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.7233688354492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8614105224609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 270, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09972341284155846, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.01805070983245969, + "grad_norm": 0.11976539343595505, + "kept_token_count": 736.525, + "learning_rate": 5.58e-06, + "loss": 0.0226, + "mgd_cos_sim": 0.9908180594444275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918193655088544, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.65332794189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.25416870117186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 280, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.10012963637709618, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.0177418552339077, + "grad_norm": 0.12123862653970718, + "kept_token_count": 736.2015625, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0222, + "mgd_cos_sim": 0.9911010682582855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008898937609046698, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.61042022705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5162048339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 290, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.10028051435947419, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.02126241158694029, + "grad_norm": 0.1223720833659172, + "kept_token_count": 735.95625, + "learning_rate": 5.98e-06, + "loss": 0.0262, + "mgd_cos_sim": 0.9901511013507843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009848912060260773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.7708297729492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6272399902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 300, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.09966151788830757, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.021744255907833576, + "grad_norm": 0.11832954734563828, + "kept_token_count": 736.784375, + "learning_rate": 6.18e-06, + "loss": 0.0263, + "mgd_cos_sim": 0.9907886505126953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009211342968046664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.8250717163086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0790222167969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09925714135169983, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.015477181691676379, + "grad_norm": 0.1270390897989273, + "kept_token_count": 737.0015625, + "learning_rate": 6.38e-06, + "loss": 0.0199, + "mgd_cos_sim": 0.9912449002265931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087550925090909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.55449371337892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9834259033203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 320, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.09963137656450272, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.02034378033131361, + "grad_norm": 0.14480793476104736, + "kept_token_count": 736.6171875, + "learning_rate": 6.58e-06, + "loss": 0.0252, + "mgd_cos_sim": 0.9902524530887604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009747557621449233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.4363525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.67455139160154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 330, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09983867332339287, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.023461760114878417, + "grad_norm": 0.12999677658081055, + "kept_token_count": 736.6140625, + "learning_rate": 6.78e-06, + "loss": 0.0282, + "mgd_cos_sim": 0.9904447317123413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009555252455174922, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.2474822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73533325195314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 340, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.10038995146751403, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.019235451798886062, + "grad_norm": 0.13168591260910034, + "kept_token_count": 735.8625, + "learning_rate": 6.98e-06, + "loss": 0.0237, + "mgd_cos_sim": 0.9911301493644714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008869844861328602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.86223754882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0200225830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 350, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.09942237660288811, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.02106786798685789, + "grad_norm": 0.10432608425617218, + "kept_token_count": 737.15, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0256, + "mgd_cos_sim": 0.990944218635559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00905578611418605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.4169952392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.99443969726565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 360, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.10031490996479989, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.018329970724880695, + "grad_norm": 0.09895569086074829, + "kept_token_count": 736.3359375, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0231, + "mgd_cos_sim": 0.9904923498630523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009507646970450877, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.99351196289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.77354431152344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 370, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.10070134922862053, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.01773229194805026, + "grad_norm": 0.11481624841690063, + "kept_token_count": 735.80625, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0227, + "mgd_cos_sim": 0.9900365889072418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009963416028767824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.95958557128907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9007232666016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 380, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09970311746001244, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.01693071089684963, + "grad_norm": 0.10140781104564667, + "kept_token_count": 736.528125, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0217, + "mgd_cos_sim": 0.9903920531272888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00960795171558857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.10277862548827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.47889404296876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 390, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.10059731900691986, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.017940949834883212, + "grad_norm": 0.1052914559841156, + "kept_token_count": 735.8890625, + "learning_rate": 7.98e-06, + "loss": 0.0225, + "mgd_cos_sim": 0.9908644378185272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009135556221008301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.61322326660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3404876708984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 400, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.10012614727020264, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.019618133921176194, + "grad_norm": 0.1066010445356369, + "kept_token_count": 736.125, + "learning_rate": 8.18e-06, + "loss": 0.0241, + "mgd_cos_sim": 0.9910522341728211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008947765920311213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.6199508666992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2288848876953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 410, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.09923971220850944, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.017801970709115267, + "grad_norm": 0.09458139538764954, + "kept_token_count": 737.08125, + "learning_rate": 8.38e-06, + "loss": 0.0224, + "mgd_cos_sim": 0.990836638212204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009163358900696039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.95084533691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.56043395996096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 420, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.10004978403449058, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.0205266984179616, + "grad_norm": 0.15006117522716522, + "kept_token_count": 736.3078125, + "learning_rate": 8.58e-06, + "loss": 0.0256, + "mgd_cos_sim": 0.9898376286029815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010162370651960373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.7867874145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3848480224609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.10083599463105201, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.017917385417968036, + "grad_norm": 0.1340678483247757, + "kept_token_count": 736.0203125, + "learning_rate": 8.78e-06, + "loss": 0.0225, + "mgd_cos_sim": 0.9907569885253906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009243009705096483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.56185455322264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.73459777832034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 440, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.09900465309619903, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.01693340754136443, + "grad_norm": 0.10578592866659164, + "kept_token_count": 737.15625, + "learning_rate": 8.98e-06, + "loss": 0.0212, + "mgd_cos_sim": 0.9914420425891877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008557952661067247, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.63002166748046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.13855895996096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.10026436746120453, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.016706290002912284, + "grad_norm": 0.11978273838758469, + "kept_token_count": 736.2171875, + "learning_rate": 9.18e-06, + "loss": 0.0211, + "mgd_cos_sim": 0.9911709129810333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008829075936228037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.54840698242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2500274658203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 460, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.0993797905743122, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.01638493426144123, + "grad_norm": 0.10819269716739655, + "kept_token_count": 737.015625, + "learning_rate": 9.38e-06, + "loss": 0.0212, + "mgd_cos_sim": 0.9904610574245453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009538936521857977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.5236053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4455963134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 470, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09940707534551621, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.018075501173734666, + "grad_norm": 0.1378508359193802, + "kept_token_count": 737.06875, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0227, + "mgd_cos_sim": 0.990665465593338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009334531240165233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.58809967041014, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3083557128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 480, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.10055216774344444, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.01764434399083257, + "grad_norm": 0.14405786991119385, + "kept_token_count": 736.10625, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0223, + "mgd_cos_sim": 0.9907578110694886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009242193400859832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.37152252197265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3224334716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 490, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09973181039094925, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.01709663113579154, + "grad_norm": 0.0977163314819336, + "kept_token_count": 736.3953125, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0214, + "mgd_cos_sim": 0.9914538323879242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008546157553792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.03612670898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5691284179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 500, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.10078497976064682, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.015062828734517097, + "grad_norm": 0.11898878961801529, + "kept_token_count": 735.734375, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9912280321121216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008771963603794574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.2158233642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.86516723632815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 510, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.09977731704711915, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.018467610329389574, + "grad_norm": 0.13303227722644806, + "kept_token_count": 736.721875, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9910239160060883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008976087160408496, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.29721069335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4712249755859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 520, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09991625919938088, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.017819690704345702, + "grad_norm": 0.13381290435791016, + "kept_token_count": 736.4515625, + "learning_rate": 1.058e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9909702420234681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009029752295464277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.9274932861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.84246826171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 530, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09935420528054237, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.019519166089594364, + "grad_norm": 0.1362682580947876, + "kept_token_count": 736.8875, + "learning_rate": 1.078e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9907317936420441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009268199373036624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.9821563720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.89117126464845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 540, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.09935320168733597, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.015150163508951663, + "grad_norm": 0.12279793620109558, + "kept_token_count": 737.215625, + "learning_rate": 1.098e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9904094159603118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009590585250407457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.37483825683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0734100341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 550, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.10007829889655113, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.01893021250143647, + "grad_norm": 0.12089148908853531, + "kept_token_count": 736.1953125, + "learning_rate": 1.118e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9910795867443085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00892042014747858, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.88778839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0740997314453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 560, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.09873342365026475, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.01878543235361576, + "grad_norm": 0.12470954656600952, + "kept_token_count": 737.421875, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9920158326625824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007984175067394971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.4844741821289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7668182373047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 570, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.09949526265263557, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.017925823759287595, + "grad_norm": 0.1269109696149826, + "kept_token_count": 736.6453125, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9914656579494476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008534339722245932, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.27830810546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6667083740234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 580, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.09979405775666236, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.01810768023133278, + "grad_norm": 0.10842419415712357, + "kept_token_count": 736.5953125, + "learning_rate": 1.178e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9907394230365754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009260580316185952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.9404754638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1865295410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 590, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1005497545003891, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.01805631220340729, + "grad_norm": 0.10570695251226425, + "kept_token_count": 736.06875, + "learning_rate": 1.198e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9916643977165223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0083356112241745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.40613250732423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2404266357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 600, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10021034479141236, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.01660893829539418, + "grad_norm": 0.1286480873823166, + "kept_token_count": 736.14375, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912905991077423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008709417004138232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.38851318359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4387969970703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 610, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.09975700303912163, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.01697754906490445, + "grad_norm": 0.10068533569574356, + "kept_token_count": 736.44375, + "learning_rate": 1.238e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9911303520202637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008869656082242727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.0881332397461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2238067626953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 620, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.10010945647954941, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.01875977125018835, + "grad_norm": 0.09104900807142258, + "kept_token_count": 736.4515625, + "learning_rate": 1.258e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9909674286842346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009032582864165306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.61785736083985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5159545898438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 630, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.10030088573694229, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.018929206114262343, + "grad_norm": 0.11753685027360916, + "kept_token_count": 736.115625, + "learning_rate": 1.278e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9914062082767486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008593807369470597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.13160552978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3420867919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 640, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.1004683367908001, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.016887463815510272, + "grad_norm": 0.10666417330503464, + "kept_token_count": 735.9265625, + "learning_rate": 1.298e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918796360492707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008120355196297169, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.32915954589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.93829040527345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 650, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.10009626001119613, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.01812918474897742, + "grad_norm": 0.13065122067928314, + "kept_token_count": 736.30625, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915803015232086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008419698104262352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.52810821533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.42787780761716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 660, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.0997716911137104, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.016747845243662594, + "grad_norm": 0.1210603192448616, + "kept_token_count": 736.7359375, + "learning_rate": 1.338e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9916299283504486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008370059356093407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.38834533691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4838409423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 670, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.09976014569401741, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.016104079876095058, + "grad_norm": 0.1300652027130127, + "kept_token_count": 736.58125, + "learning_rate": 1.358e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9916811406612396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008318862784653902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.83466644287108, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5296600341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 680, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.09951581954956054, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.014854733645915986, + "grad_norm": 0.09186125546693802, + "kept_token_count": 736.921875, + "learning_rate": 1.378e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9918859302997589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008114060293883086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.9491180419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.575244140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 690, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.10037793591618538, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.016554967407137154, + "grad_norm": 0.14520689845085144, + "kept_token_count": 736.3546875, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9916480600833892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008351948112249374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.1015182495117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1732940673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 700, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.09914035350084305, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.016681421361863614, + "grad_norm": 0.11304144561290741, + "kept_token_count": 737.4453125, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9911247313022613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008875266462564469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.20130157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3350555419922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 710, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.10000096186995507, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.01604887517169118, + "grad_norm": 0.10113412886857986, + "kept_token_count": 736.2171875, + "learning_rate": 1.438e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910119593143463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008988054655492305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.02301788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.54082946777345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 720, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.10003671944141387, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.016189651750028133, + "grad_norm": 0.09254481643438339, + "kept_token_count": 736.4828125, + "learning_rate": 1.458e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913931310176849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008606863021850587, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.41010131835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.81206970214845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 730, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.10064059868454933, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.018629332538694143, + "grad_norm": 0.11100048571825027, + "kept_token_count": 735.8625, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9912373423576355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008762639109045267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.52033538818358, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73011169433596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 740, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09958676248788834, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.018697952199727298, + "grad_norm": 0.13826106488704681, + "kept_token_count": 736.8953125, + "learning_rate": 1.498e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9912423491477966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00875767944380641, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.5636428833008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.583984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 750, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.0999738983809948, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.017898683622479438, + "grad_norm": 0.14949294924736023, + "kept_token_count": 736.340625, + "learning_rate": 1.518e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9917865097522736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0082134947180748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.2634033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9879486083984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 760, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.09981243312358856, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.02037272658199072, + "grad_norm": 0.15102887153625488, + "kept_token_count": 736.3984375, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9908171772956849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918280826881528, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.64663543701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0791839599609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 770, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.10012710392475128, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.017416724376380442, + "grad_norm": 0.12118300795555115, + "kept_token_count": 736.1671875, + "learning_rate": 1.558e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915714681148529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008428527787327767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.7159210205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7139495849609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 780, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.09973175972700118, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.015414332412183285, + "grad_norm": 0.1227661594748497, + "kept_token_count": 736.2484375, + "learning_rate": 1.578e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9919974446296692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008002542983740568, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.83729705810546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7450286865234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 790, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.0993483081459999, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.01777557535097003, + "grad_norm": 0.12679772078990936, + "kept_token_count": 736.6953125, + "learning_rate": 1.598e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9911988019943238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008801191300153732, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.74476776123046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.34343872070315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 800, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.0998285785317421, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.014140446949750186, + "grad_norm": 0.09576638787984848, + "kept_token_count": 736.49375, + "learning_rate": 1.618e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.991389524936676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008610486704856157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.28384399414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8094421386719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 810, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09996364563703537, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.016697122342884542, + "grad_norm": 0.12903039157390594, + "kept_token_count": 736.2953125, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9914243161678314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008575688768178224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.34282531738282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.09429321289065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 820, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.10001128688454627, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.016198556497693063, + "grad_norm": 0.11097489297389984, + "kept_token_count": 736.1515625, + "learning_rate": 1.658e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9915563583374023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008443635236471892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.26085815429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7219299316406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 830, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.09985358789563178, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.018463446106761695, + "grad_norm": 0.1168319582939148, + "kept_token_count": 736.596875, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.991520220041275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00847977316007018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.4137924194336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.06837158203126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 840, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.09962129071354867, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.018385927379131316, + "grad_norm": 0.1216987743973732, + "kept_token_count": 736.68125, + "learning_rate": 1.698e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9908582389354705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009141768887639045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.34853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1119049072266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 850, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.09975438266992569, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.01790990522131324, + "grad_norm": 0.1288978010416031, + "kept_token_count": 736.9046875, + "learning_rate": 1.718e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913394331932068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0086605753749609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.93800506591796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.41443786621096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 860, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.1005734957754612, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.0176123715005815, + "grad_norm": 0.12454768270254135, + "kept_token_count": 735.6625, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9905102550983429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009489736799150706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.00812530517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3312652587891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 870, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.10008746162056922, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.01822800440713763, + "grad_norm": 0.10062307119369507, + "kept_token_count": 736.6203125, + "learning_rate": 1.758e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9907923817634583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009207623358815909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.7578887939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9511779785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 880, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.10055599510669708, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.01665054364129901, + "grad_norm": 0.12177368998527527, + "kept_token_count": 735.765625, + "learning_rate": 1.778e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9914739966392517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852600708603859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.36559600830077, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.00425720214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 890, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.09999357089400292, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.018511239904910326, + "grad_norm": 0.09459419548511505, + "kept_token_count": 736.3578125, + "learning_rate": 1.798e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9914514720439911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008548529073596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.47552490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8278472900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 900, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.10014134421944618, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.01755330553278327, + "grad_norm": 0.12321103364229202, + "kept_token_count": 736.15625, + "learning_rate": 1.818e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9909531772136688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009046830609440804, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.7320541381836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.34531555175784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 910, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.09906976893544198, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.015752295218408107, + "grad_norm": 0.12000124156475067, + "kept_token_count": 737.309375, + "learning_rate": 1.838e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9916189253330231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008381052874028683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.44276580810546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9455261230469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 920, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.09991866499185562, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.018705749325454234, + "grad_norm": 0.11857365816831589, + "kept_token_count": 736.64375, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9910420656204224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008957943506538868, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.74480133056642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9513916015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 930, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.10001512765884399, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.017535715736448763, + "grad_norm": 0.1039995327591896, + "kept_token_count": 736.434375, + "learning_rate": 1.878e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9904098510742188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00959015116095543, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.88767700195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9598663330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 940, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09981609657406806, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.016502436343580486, + "grad_norm": 0.13716939091682434, + "kept_token_count": 736.9578125, + "learning_rate": 1.898e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9918734729290009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008126529306173325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.9645736694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.36011962890626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 950, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.10075157284736633, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.017337518092244864, + "grad_norm": 0.12811565399169922, + "kept_token_count": 736.084375, + "learning_rate": 1.918e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9914735794067383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008526426739990711, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.06367034912108, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9185333251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 960, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.10000531449913978, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.017770913429558277, + "grad_norm": 0.12776786088943481, + "kept_token_count": 736.440625, + "learning_rate": 1.938e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9910569727420807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008943028561770917, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.52740325927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.525244140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10085626617074013, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.014635485876351596, + "grad_norm": 0.0989716500043869, + "kept_token_count": 735.565625, + "learning_rate": 1.958e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9920821249485016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0079178711399436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.7938430786133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.74671325683596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 980, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.09980271682143212, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.02125456966459751, + "grad_norm": 0.13963866233825684, + "kept_token_count": 736.9203125, + "learning_rate": 1.978e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9911113739013672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008888635039329528, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.84815979003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1936004638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 990, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.10031667277216912, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.0163023485802114, + "grad_norm": 0.12437321245670319, + "kept_token_count": 736.253125, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.991327303647995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008672708552330733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.7868667602539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4775909423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1000, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.10001121461391449, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.017158953472971916, + "grad_norm": 0.1440173089504242, + "kept_token_count": 736.440625, + "learning_rate": 2.018e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9909565448760986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009043449349701405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.03980407714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.72284545898435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1010, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.10007239505648613, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.021448755729943515, + "grad_norm": 0.15623342990875244, + "kept_token_count": 736.3265625, + "learning_rate": 2.038e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9909003615379334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00909965019673109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.31566925048827, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.85971984863284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.10008150115609168, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.016338780149817468, + "grad_norm": 0.1357824206352234, + "kept_token_count": 736.5328125, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.991439265012741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008560737501829863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.37588500976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4608093261719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1030, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.09983802065253258, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.015365237928926944, + "grad_norm": 0.1173689141869545, + "kept_token_count": 736.6984375, + "learning_rate": 2.078e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9904743790626526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009525615628808737, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.58472442626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.66844482421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1040, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.10055146664381027, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.016601147968322037, + "grad_norm": 0.1368919163942337, + "kept_token_count": 736.3625, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9910662710666657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008933732472360133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.81488647460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.529736328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1050, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.10030929297208786, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.014289958216249943, + "grad_norm": 0.12890781462192535, + "kept_token_count": 736.0671875, + "learning_rate": 2.118e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9909204483032227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009079540148377419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.95196990966798, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6916442871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1060, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.10063716173171997, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.017282093595713376, + "grad_norm": 0.12827515602111816, + "kept_token_count": 736.084375, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909970819950104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900291819125414, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.7039596557617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.64764099121095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1070, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.10005562901496887, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.01851187478750944, + "grad_norm": 0.16248232126235962, + "kept_token_count": 736.4609375, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9909257650375366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009074240084737539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.3550048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.47548217773436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1080, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09939933717250823, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.015849738847464323, + "grad_norm": 0.14303627610206604, + "kept_token_count": 736.9078125, + "learning_rate": 2.178e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9915820777416229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008417918719351292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.3449691772461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6510314941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1090, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.0998035304248333, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.016219654586166143, + "grad_norm": 0.10981979221105576, + "kept_token_count": 736.55, + "learning_rate": 2.198e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9913254380226135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008674564026296138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.1197937011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4594360351563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1100, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.10027582049369813, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.015166979096829892, + "grad_norm": 0.13102377951145172, + "kept_token_count": 736.271875, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9910475552082062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00895244050770998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.6491485595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.08006591796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1110, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.09992947280406952, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.015358926635235548, + "grad_norm": 0.09523575752973557, + "kept_token_count": 736.6296875, + "learning_rate": 2.238e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9914029479026795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008597056660801173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.81605377197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.72052307128905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1120, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.09933049976825714, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.018185100238770246, + "grad_norm": 0.15178914368152618, + "kept_token_count": 737.09375, + "learning_rate": 2.258e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9913301885128021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008669811021536588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.76869506835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3427398681641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1130, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.09944036453962327, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.018660357780754568, + "grad_norm": 0.14189158380031586, + "kept_token_count": 737.125, + "learning_rate": 2.278e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9905837953090668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009416201990097762, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.1609313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.77106018066405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1140, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.09966066777706147, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.014447907079011202, + "grad_norm": 0.11948457360267639, + "kept_token_count": 736.584375, + "learning_rate": 2.298e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9917062401771546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00829374473541975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.35995635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.78001098632814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1150, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.10054119974374771, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.017839243356138467, + "grad_norm": 0.1345304399728775, + "kept_token_count": 736.03125, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913783311843872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008621673472225666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.3630798339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8033782958984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1160, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.1005743570625782, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.019132527522742748, + "grad_norm": 0.1142367348074913, + "kept_token_count": 735.68125, + "learning_rate": 2.338e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9909660995006562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009033896122127771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.5299041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8989685058594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1170, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.10011267513036728, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.018235685210675, + "grad_norm": 0.14310836791992188, + "kept_token_count": 736.49375, + "learning_rate": 2.358e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9904176831245423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009582322835922242, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.17793884277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.14564514160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1180, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.10011869594454766, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.015561853628605604, + "grad_norm": 0.15167713165283203, + "kept_token_count": 736.41875, + "learning_rate": 2.378e-05, + "loss": 0.02, + "mgd_cos_sim": 0.991200965642929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008799036033451557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.5196990966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1514434814453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1190, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.09913590401411057, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.019650013744831087, + "grad_norm": 0.14897117018699646, + "kept_token_count": 737.3484375, + "learning_rate": 2.398e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9900890767574311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009910925477743148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.81410064697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47088623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1200, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.09978258311748504, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.01580818146467209, + "grad_norm": 0.10314518213272095, + "kept_token_count": 736.29375, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9913116872310639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008688303641974926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.92119293212892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.25892333984376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1210, + "valid_token_count": 817.9078125 + }, + { + "actual_token_mask_ratio": 0.10072154700756072, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.016018849331885576, + "grad_norm": 0.20188695192337036, + "kept_token_count": 735.83125, + "learning_rate": 2.438e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9909595847129822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040432143956422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.109814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9721954345703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1220, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09980099275708199, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.017784419190138578, + "grad_norm": 0.1710938960313797, + "kept_token_count": 736.1421875, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9908859610557557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00911404239013791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.05152893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.68175048828124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1230, + "valid_token_count": 817.75625 + }, + { + "actual_token_mask_ratio": 0.10049119293689728, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.016760175675153734, + "grad_norm": 0.13561855256557465, + "kept_token_count": 736.0296875, + "learning_rate": 2.478e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9911050260066986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008894987404346466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.59134368896486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.99155578613284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1240, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10012766271829605, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.015989054273813964, + "grad_norm": 0.1835946887731552, + "kept_token_count": 736.253125, + "learning_rate": 2.498e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910041868686676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00899581704288721, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.9240753173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9981170654297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1250, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.09940927028656006, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.017081534676253795, + "grad_norm": 0.12124525755643845, + "kept_token_count": 737.053125, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9915475904941559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00845239320769906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.58414306640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2983367919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1260, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.09947839975357056, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.015845964662730694, + "grad_norm": 0.1120937317609787, + "kept_token_count": 736.8296875, + "learning_rate": 2.538e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9913357377052308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008664276637136935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.6863250732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91168212890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1270, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.10044393241405487, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.017595366202294828, + "grad_norm": 0.11958418041467667, + "kept_token_count": 736.096875, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.991491025686264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008508974127471447, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.59011993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.82171325683595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1280, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.0999143086373806, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.01706320820376277, + "grad_norm": 0.11268851906061172, + "kept_token_count": 736.1921875, + "learning_rate": 2.578e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9912166953086853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00878330310806632, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.7647445678711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1229522705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1290, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.10032324865460396, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.0164445873349905, + "grad_norm": 0.13028059899806976, + "kept_token_count": 736.025, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9909766554832459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009023351967334748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.43533172607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1859161376953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1300, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.10011815726757049, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.015228288620710373, + "grad_norm": 0.10474733263254166, + "kept_token_count": 736.409375, + "learning_rate": 2.618e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9909846782684326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009015326201915742, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.56178588867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.80913696289065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.10016532391309738, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.014550419058650732, + "grad_norm": 0.13615041971206665, + "kept_token_count": 736.2859375, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9909129559993743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009087064117193223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 206.85625457763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1759521484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1320, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10028782859444618, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.019689288362860678, + "grad_norm": 0.13791503012180328, + "kept_token_count": 736.3609375, + "learning_rate": 2.658e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9903671741485596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00963281812146306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.35960998535157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.17900390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1330, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.10058225467801094, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.018357645068317652, + "grad_norm": 0.1556171327829361, + "kept_token_count": 736.259375, + "learning_rate": 2.678e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9899641275405884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010035865195095538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.87274017333985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8710968017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1340, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.10015127211809158, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.016939748264849185, + "grad_norm": 0.1320413202047348, + "kept_token_count": 736.3265625, + "learning_rate": 2.698e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9908671975135803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009132792335003614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.31682281494142, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1886413574219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1350, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.09998543262481689, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.017050577513873576, + "grad_norm": 0.12157519161701202, + "kept_token_count": 736.4125, + "learning_rate": 2.718e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909627377986908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009037247952073812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.30588989257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.87720947265626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1360, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.09991183653473854, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.018213045410811902, + "grad_norm": 0.14183634519577026, + "kept_token_count": 736.634375, + "learning_rate": 2.738e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.990739232301712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009260751120746136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.51978759765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.21600341796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1370, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.09976698756217957, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.015592510905116796, + "grad_norm": 0.10158330202102661, + "kept_token_count": 736.8421875, + "learning_rate": 2.758e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9917796134948731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00822038296610117, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.75523986816407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.51942749023436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1380, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.09895242154598236, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.01684054723009467, + "grad_norm": 0.1124860942363739, + "kept_token_count": 737.18125, + "learning_rate": 2.778e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9908658862113953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009134112298488617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.65245513916017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.102685546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1390, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.09959316551685334, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.01647953875362873, + "grad_norm": 0.13448818027973175, + "kept_token_count": 736.9828125, + "learning_rate": 2.798e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9912562251091004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008743777219206095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.1770004272461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.67564697265624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1400, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.10027063637971878, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.016756384540349244, + "grad_norm": 0.18678680062294006, + "kept_token_count": 736.2890625, + "learning_rate": 2.818e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9899789273738862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010021086316555738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.01524353027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5993927001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1410, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.09963446035981179, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.01727525545284152, + "grad_norm": 0.10778346657752991, + "kept_token_count": 736.6546875, + "learning_rate": 2.838e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9901691615581513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009830842446535825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.61058654785157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8338104248047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1420, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.0995584212243557, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.01638247733935714, + "grad_norm": 0.11082155257463455, + "kept_token_count": 736.7109375, + "learning_rate": 2.858e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9915382444858551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00846175691112876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.26932067871093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2796142578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.0997086375951767, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.015451480355113744, + "grad_norm": 0.10700030624866486, + "kept_token_count": 736.5890625, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9909574151039123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009042603056877851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.9565002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.22307739257815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1440, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.100174942612648, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.015522562619298696, + "grad_norm": 0.10682902485132217, + "kept_token_count": 736.35625, + "learning_rate": 2.898e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9913280963897705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008671907614916563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.54954986572267, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5776672363281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1450, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.10073782876133919, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.021279547922313215, + "grad_norm": 0.13553445041179657, + "kept_token_count": 735.6953125, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.990392655134201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009607329685240983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.21471405029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.72130432128904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1460, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.10009672343730927, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.01709520062431693, + "grad_norm": 0.09564915299415588, + "kept_token_count": 736.409375, + "learning_rate": 2.938e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909543573856354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009045644570142031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.17029571533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4221252441406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1470, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09981521517038346, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.01855683671310544, + "grad_norm": 0.11719273775815964, + "kept_token_count": 736.775, + "learning_rate": 2.958e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9910685002803803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008931500371545553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.0270736694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.18450317382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1480, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.09912176653742791, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.01764199249446392, + "grad_norm": 0.15928415954113007, + "kept_token_count": 737.1265625, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908821821212769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00911781694740057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.68106384277343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0794677734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1490, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.09969770014286042, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.017065525986254215, + "grad_norm": 0.14774316549301147, + "kept_token_count": 736.8640625, + "learning_rate": 2.998e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9906425714492798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009357428923249245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.58230438232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7793914794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1500, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.10039280652999878, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.01782135320827365, + "grad_norm": 0.12622514367103577, + "kept_token_count": 735.7078125, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9903309285640717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009669059049338102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.10149688720702, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0083923339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1510, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.09998151287436485, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.015266004391014576, + "grad_norm": 0.1321694403886795, + "kept_token_count": 736.359375, + "learning_rate": 2.999996710133069e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9905031859874726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009496815223246812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.39065551757812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4342926025391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1520, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.10011783540248871, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.017770847585052253, + "grad_norm": 0.12436346709728241, + "kept_token_count": 736.565625, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9915058553218842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008494133315980434, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.49208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4151245117188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1530, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.09992623105645179, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.018558734841644765, + "grad_norm": 0.15415865182876587, + "kept_token_count": 736.5609375, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9909426629543304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009057336207479238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.92267608642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9290313720703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1540, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.10032929331064225, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.01835023211315274, + "grad_norm": 0.1396421492099762, + "kept_token_count": 736.590625, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9909002780914307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00909972907975316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.69564361572264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.07220764160155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1550, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.10030737817287445, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.018772224709391595, + "grad_norm": 0.09987959265708923, + "kept_token_count": 736.00625, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9904963850975037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009503617323935032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.00672302246093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0121246337891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1560, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.10045867785811424, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.01940910778939724, + "grad_norm": 0.12420545518398285, + "kept_token_count": 736.1390625, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9905740261077881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00942597510293126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.2998245239258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.37334899902345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1570, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.09967277720570564, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.01777060218155384, + "grad_norm": 0.16196773946285248, + "kept_token_count": 736.5046875, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9901202380657196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00987975373864174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.00750732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.69195861816405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1580, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.09994393661618232, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.018319414369761942, + "grad_norm": 0.11864075064659119, + "kept_token_count": 736.684375, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9898526966571808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010147301480174065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.0223602294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9209045410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1590, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.10023851469159126, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.018906875047832727, + "grad_norm": 0.11084339022636414, + "kept_token_count": 736.3265625, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9907730460166931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009226940013468266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.863916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4814483642578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1600, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.09980039075016975, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.018196386005729436, + "grad_norm": 0.11709818989038467, + "kept_token_count": 736.59375, + "learning_rate": 2.999891727276827e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9904141008853913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009585888870060445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.1425354003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1975433349609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1610, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.09945538491010666, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.016033842973411084, + "grad_norm": 0.10805010050535202, + "kept_token_count": 736.565625, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9905992388725281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009400764014571905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.37068939208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.41615600585936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1620, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.10036118850111961, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.02023873580619693, + "grad_norm": 0.12030686438083649, + "kept_token_count": 735.953125, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9904880344867706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009511968120932579, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.67837982177736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7850830078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.1007456786930561, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.01605241307988763, + "grad_norm": 0.16011732816696167, + "kept_token_count": 735.9140625, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9905239522457123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009476046916097403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.1811050415039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.18677673339846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.10044624954462052, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.017290350608527662, + "grad_norm": 0.1285509467124939, + "kept_token_count": 735.9640625, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9898428916931152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010157120786607266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.31137237548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.252490234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.09979721307754516, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.016185152158141137, + "grad_norm": 0.12669023871421814, + "kept_token_count": 736.58125, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9902152955532074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009784718696027993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.83460083007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3219329833984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1660, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.10070479586720467, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.01830313354730606, + "grad_norm": 0.11375865340232849, + "kept_token_count": 736.184375, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9909106194972992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009089376125484706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.91607208251952, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.06622619628905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1670, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.09944943115115165, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.01852258322760463, + "grad_norm": 0.15737144649028778, + "kept_token_count": 736.9640625, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9905512511730195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009448736067861318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.08274688720704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.83470153808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1680, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10060483366250991, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.020517214108258484, + "grad_norm": 0.14771655201911926, + "kept_token_count": 735.8171875, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9904589116573334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009541097842156888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.39529724121093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.84791259765626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.09947762116789818, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.020712577272206545, + "grad_norm": 0.12262618541717529, + "kept_token_count": 736.9203125, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9906952381134033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00930475303903222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.53316802978514, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.72771911621095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.09994832798838615, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.018289645295590164, + "grad_norm": 0.14171430468559265, + "kept_token_count": 736.4546875, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908609688282013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009139026887714863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.8301010131836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0907501220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1710, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.09967975616455078, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.0180697126314044, + "grad_norm": 0.1413324624300003, + "kept_token_count": 736.56875, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9909420192241669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00905798077583313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.13038330078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9696929931641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1720, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.10083996877074242, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.01732952082529664, + "grad_norm": 0.13874110579490662, + "kept_token_count": 735.65625, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9909054577350617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009094545431435109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.0022201538086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2239654541016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1730, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.09986376091837883, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.01726224645972252, + "grad_norm": 0.11366185545921326, + "kept_token_count": 736.571875, + "learning_rate": 2.999479474546582e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9905630350112915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009436962474137545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.96442260742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1351318359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1740, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.1001538135111332, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.01919921236112714, + "grad_norm": 0.1477798968553543, + "kept_token_count": 736.33125, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9903632938861847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009636704716831446, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.27178039550782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5373291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1750, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.10057792142033577, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.01666814535856247, + "grad_norm": 0.1456276774406433, + "kept_token_count": 736.13125, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9911898732185364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008810110576450824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.0641845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.95383605957034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1760, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.09982823207974434, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.01957969954237342, + "grad_norm": 0.15319512784481049, + "kept_token_count": 736.2953125, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9911669254302978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008833088632673024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.81056213378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8240661621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1770, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.10008014813065529, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.017258144728839398, + "grad_norm": 0.11678777635097504, + "kept_token_count": 736.4171875, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.990865272283554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009134725574404002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.56500244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.75726318359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1780, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.09962406605482102, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.018833988159894944, + "grad_norm": 0.18333841860294342, + "kept_token_count": 736.6609375, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9915172219276428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008482776861637831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.56079559326173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.99761962890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1790, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09964423030614852, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.01929902667179704, + "grad_norm": 0.1571682244539261, + "kept_token_count": 736.7375, + "learning_rate": 2.999185343831476e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9905149638652802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009485044423490763, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.50465545654296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.78289489746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1800, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.0999940313398838, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.01659831264987588, + "grad_norm": 0.13749869167804718, + "kept_token_count": 736.3578125, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9906964957714081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009303530771285295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.4293426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.85398254394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1810, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.1002627469599247, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.018251369707286358, + "grad_norm": 0.11903063207864761, + "kept_token_count": 736.2453125, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9908069372177124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009193055238574744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.14530029296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0202331542969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1820, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09972397908568383, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.016380561143159868, + "grad_norm": 0.14488168060779572, + "kept_token_count": 737.0, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.990862500667572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009137515537440776, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.6258529663086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.50242004394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1830, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.10042900294065475, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.019931568671017884, + "grad_norm": 0.12770448625087738, + "kept_token_count": 736.178125, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.025, + "mgd_cos_sim": 0.989908903837204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010091092810034752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.4961166381836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0376770019531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1840, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.10055361539125443, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.018202314991503953, + "grad_norm": 0.14725783467292786, + "kept_token_count": 735.79375, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9908384859561921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009161515161395073, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.21180267333983, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.37114868164065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1850, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.10015532672405243, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.018815301079303026, + "grad_norm": 0.11783064156770706, + "kept_token_count": 736.121875, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9906147181987762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938529083505273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.3613021850586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7626617431641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1860, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.09999408796429635, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.01761830262839794, + "grad_norm": 0.10226020216941833, + "kept_token_count": 736.5, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9909534335136414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009046563226729632, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.23121490478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.56850280761716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1870, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.10026033967733383, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.01714507834985852, + "grad_norm": 0.13613678514957428, + "kept_token_count": 736.3703125, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9912177205085755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008782292529940606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.88869934082032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.39353942871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1880, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.10019687935709953, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.019846886675804854, + "grad_norm": 0.12697383761405945, + "kept_token_count": 736.4828125, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.990849643945694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009150347672402859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.10853424072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7164733886719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1890, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.09931955710053444, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.01757325464859605, + "grad_norm": 0.14452584087848663, + "kept_token_count": 737.08125, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907111287117004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009288880228996276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.4461868286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.61760864257815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1900, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.09978452175855637, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.01836781557649374, + "grad_norm": 0.139029860496521, + "kept_token_count": 736.6421875, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9901869595050812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00981304319575429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.95755004882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91595458984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.10014656782150269, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.017425073590129615, + "grad_norm": 0.10983099043369293, + "kept_token_count": 736.1625, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9907431423664093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009256852697581053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.03170623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.46556701660154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.10025900825858117, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.01792766572907567, + "grad_norm": 0.11884074658155441, + "kept_token_count": 735.921875, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9907257378101348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00927426852285862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.77509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1848205566406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1930, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.09983486458659172, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.019453131686896087, + "grad_norm": 0.1072547435760498, + "kept_token_count": 736.5890625, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.990792739391327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009207261726260185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.4313186645508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.13173522949216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.09949752911925316, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.017517619673162697, + "grad_norm": 0.16215667128562927, + "kept_token_count": 736.8453125, + "learning_rate": 2.998163143144622e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910522401332855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008947763685137033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.47498016357423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3936431884766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1950, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.1000655621290207, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.01685465034097433, + "grad_norm": 0.12463336437940598, + "kept_token_count": 736.4609375, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917356848716736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008264320064336061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.90010528564454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6941345214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1960, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.1002211518585682, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.019660894852131607, + "grad_norm": 0.1914091855287552, + "kept_token_count": 736.309375, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9904667794704437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009533224254846573, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.5415252685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7077331542969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1970, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.10022147446870804, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.018318928219377993, + "grad_norm": 0.12530554831027985, + "kept_token_count": 736.46875, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908932209014892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009106771554797888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.63280487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8035095214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1980, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.09950810074806213, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.020192477293312548, + "grad_norm": 0.11102958768606186, + "kept_token_count": 736.8765625, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.98976349234581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010236504953354597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.86929626464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.02378845214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 1990, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.09961663633584976, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.016477326676249505, + "grad_norm": 0.12390560656785965, + "kept_token_count": 736.75, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9908448278903961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009155154507607222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.75421752929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.95305786132815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2000, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09999606236815453, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.017533013224601747, + "grad_norm": 0.12817980349063873, + "kept_token_count": 736.2453125, + "learning_rate": 2.997639560260358e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910187780857086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008981232345104218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.69719848632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.83433227539064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.10120408684015274, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.019202546495944262, + "grad_norm": 0.12772314250469208, + "kept_token_count": 735.3828125, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9910858869552612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00891410345211625, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.30170440673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.67662658691404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2020, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.09974826946854591, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.018442586343735457, + "grad_norm": 0.1627555936574936, + "kept_token_count": 736.5203125, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9907789945602417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009220988489687443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.40426788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6264129638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2030, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09985542222857476, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.017478112131357193, + "grad_norm": 0.11449676007032394, + "kept_token_count": 736.4578125, + "learning_rate": 2.997353200422555e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9909597098827362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040302503854037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.13485870361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4980865478516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2040, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09926525428891182, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.01917099514976144, + "grad_norm": 0.15203219652175903, + "kept_token_count": 737.01875, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9912765026092529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008723485469818115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.14677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1405487060547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.1002037726342678, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.018355353735387325, + "grad_norm": 0.12112722545862198, + "kept_token_count": 736.2703125, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908918797969818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009108113683760166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.87416229248046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.78636779785154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2060, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10005119740962982, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.020900464802980422, + "grad_norm": 0.13587063550949097, + "kept_token_count": 736.259375, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9899413824081421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010058632399886847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.41551055908204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.39026794433596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2070, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.10009215399622917, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.017160562425851823, + "grad_norm": 0.18280243873596191, + "kept_token_count": 736.4234375, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9899292409420013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010070746298879385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.399560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3916534423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2080, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09972216337919235, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.01916950074955821, + "grad_norm": 0.1436549574136734, + "kept_token_count": 736.41875, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9911864638328552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008813545387238265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.39514007568357, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.63010864257814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2090, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.0991067111492157, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.01847229674458504, + "grad_norm": 0.13940173387527466, + "kept_token_count": 737.346875, + "learning_rate": 2.996731359613498e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9909623622894287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009037636406719684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.2725570678711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7221313476563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2100, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.1001454196870327, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.016270521190017463, + "grad_norm": 0.1416521519422531, + "kept_token_count": 736.3921875, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9900170087814331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009982990194112062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.8763854980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.75372009277345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2110, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.09996626153588295, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.017281046696007253, + "grad_norm": 0.14698286354541779, + "kept_token_count": 736.55, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9911497890949249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00885019265115261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.018896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0714904785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2120, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.10063785687088966, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.021087145898491144, + "grad_norm": 0.12120052427053452, + "kept_token_count": 736.0890625, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9902770519256592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00972294807434082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.52569732666015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0398834228516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2130, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.10049934685230255, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.016181548591703176, + "grad_norm": 0.13367505371570587, + "kept_token_count": 735.8640625, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.991262036561966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008737967442721128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.32295837402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.64576721191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.0997024692595005, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.017049525491893293, + "grad_norm": 0.11444870382547379, + "kept_token_count": 736.84375, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.990996515750885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009003481641411782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.2972213745117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.46836547851564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2150, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.09990911930799484, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.016508883610367776, + "grad_norm": 0.1213068887591362, + "kept_token_count": 736.621875, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.990696930885315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009303071908652783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.31289520263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9424072265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2160, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09989076480269432, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.015282315481454135, + "grad_norm": 0.15647947788238525, + "kept_token_count": 736.3890625, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9915992736816406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008400715608149766, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.1827194213867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1195404052734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2170, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09961858615279198, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.017543526459485294, + "grad_norm": 0.12603409588336945, + "kept_token_count": 736.840625, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9901106357574463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009889362286776305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.97909088134764, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0518493652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2180, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.10078794360160828, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.016255626827478407, + "grad_norm": 0.11648579686880112, + "kept_token_count": 735.6140625, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9911144137382507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00888559389859438, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.96203002929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.89226989746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.10002860128879547, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.021254288870841263, + "grad_norm": 0.17354562878608704, + "kept_token_count": 736.6578125, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9913179278373718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008682067785412074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.5121368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.71933898925784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2200, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.09970399439334869, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.019283611327409744, + "grad_norm": 0.146134153008461, + "kept_token_count": 736.975, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9910920679569244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00890793427824974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.77072296142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.90391540527344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2210, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.1001158095896244, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.019434863701462746, + "grad_norm": 0.13018088042736053, + "kept_token_count": 736.50625, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9913711607456207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008628845307976007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.6512939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.33359680175784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2220, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.10002177357673644, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.017332529462873936, + "grad_norm": 0.12761199474334717, + "kept_token_count": 736.6234375, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9910427570343018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00895725516602397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.20286712646487, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98516235351565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2230, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.09990630373358726, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.01775849061086774, + "grad_norm": 0.15589959919452667, + "kept_token_count": 736.7203125, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9909671545028687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009032849874347448, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.85325927734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5562683105469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2240, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.09993293806910515, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.01832909919321537, + "grad_norm": 0.13759447634220123, + "kept_token_count": 736.4921875, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9904491543769837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009550850000232458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.53799133300782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.25401611328124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.0999522402882576, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.018849281407892705, + "grad_norm": 0.11938145756721497, + "kept_token_count": 736.45625, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9908980906009675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00910191535949707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.60935974121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2616943359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2260, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.09988829791545868, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.017624924518167974, + "grad_norm": 0.1140182688832283, + "kept_token_count": 736.371875, + "learning_rate": 2.994614030404228e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9912454426288605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00875456165522337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.39615020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0650665283203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2270, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10011873468756675, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.01542996745556593, + "grad_norm": 0.11246185004711151, + "kept_token_count": 736.328125, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9909807324409485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009019279852509499, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.61851196289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5624298095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2280, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10042538568377495, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.018050621543079615, + "grad_norm": 0.10740195959806442, + "kept_token_count": 736.0359375, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.991122019290924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008877983409911394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.76146850585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.69439086914065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.10030392035841942, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.01782594816759229, + "grad_norm": 0.1474238783121109, + "kept_token_count": 735.8953125, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9906583070755005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009341679513454437, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.45067901611327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.10472106933594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2300, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.09990146532654762, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.019449250772595406, + "grad_norm": 0.16838037967681885, + "kept_token_count": 736.803125, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9902776896953582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009722313191741706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.4341567993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35747680664065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2310, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.10049401000142097, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.018264382891356946, + "grad_norm": 0.15922461450099945, + "kept_token_count": 736.0234375, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9901720345020294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009827965870499611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.00216674804688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0445953369141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09893399775028229, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.020065687689930202, + "grad_norm": 0.1131172776222229, + "kept_token_count": 737.2421875, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9905773341655731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00942265558987856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.32088623046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3840789794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2330, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.09975698217749596, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.016939742676913738, + "grad_norm": 0.12529124319553375, + "kept_token_count": 736.853125, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.990062814950943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009937203023582697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.509375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2806121826172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.09999274760484696, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.02137106712907553, + "grad_norm": 0.10206542164087296, + "kept_token_count": 736.240625, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9903180956840515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009681909531354903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.1671829223633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.79808044433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2350, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.09928537383675576, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.01886974275112152, + "grad_norm": 0.12743186950683594, + "kept_token_count": 737.0078125, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9901518642902374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009848133381456136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.48018341064454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1729736328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10050539746880531, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.017684451770037413, + "grad_norm": 0.13039684295654297, + "kept_token_count": 736.009375, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9904919803142548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009508035238832235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.28192443847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.97765197753904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2370, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10010195225477218, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.01735407719388604, + "grad_norm": 0.13703839480876923, + "kept_token_count": 736.475, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915882110595703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008411779999732971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.5442749023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9217803955078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2380, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.10048365294933319, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.018358083628118037, + "grad_norm": 0.13571122288703918, + "kept_token_count": 736.428125, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9912094652652741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00879054656252265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.4097961425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9436065673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2390, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.09999551549553871, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.019006378855556248, + "grad_norm": 0.12784913182258606, + "kept_token_count": 736.7234375, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9903790891170502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009620915539562703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.90340423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.56215209960936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2400, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.1004734568297863, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.016181099973618985, + "grad_norm": 0.1413181722164154, + "kept_token_count": 736.0953125, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.991034871339798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008965131733566523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.2912963867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.544775390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2410, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.10028636679053307, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.022172882035374642, + "grad_norm": 0.11642977595329285, + "kept_token_count": 736.01875, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0268, + "mgd_cos_sim": 0.99073166847229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009268349967896938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.47658081054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2922027587891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2420, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.10046230405569076, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.018551001138985158, + "grad_norm": 0.15795283019542694, + "kept_token_count": 735.8296875, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9903896272182464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009610362816601992, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.0046859741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1117340087891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2430, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.1004238747060299, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.01922211581841111, + "grad_norm": 0.11278670281171799, + "kept_token_count": 736.28125, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9912720382213592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008727951906621456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.01527099609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.78675537109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2440, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.09929081201553344, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.019070194754749536, + "grad_norm": 0.13162758946418762, + "kept_token_count": 737.2859375, + "learning_rate": 2.991800117455e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9906499564647675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00935004474595189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.6120880126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1442535400391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2450, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.10027999728918076, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.016693555936217307, + "grad_norm": 0.12482192367315292, + "kept_token_count": 736.6640625, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9903179109096527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009682082198560238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.3956024169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1865234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2460, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.10041233748197556, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.01931286547333002, + "grad_norm": 0.1431276947259903, + "kept_token_count": 736.578125, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.990896213054657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00910377074033022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.87757873535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2440551757812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2470, + "valid_token_count": 818.7953125 + }, + { + "actual_token_mask_ratio": 0.09931996688246728, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.0163878271356225, + "grad_norm": 0.10299217700958252, + "kept_token_count": 737.0234375, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9910063862800598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008993599843233824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.20420837402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.23106994628904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2480, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.10006580352783204, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.021068834140896796, + "grad_norm": 0.13725058734416962, + "kept_token_count": 736.565625, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.991496330499649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008503654226660728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.6911895751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6678161621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2490, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09985386431217194, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.018781109340488912, + "grad_norm": 0.13542737066745758, + "kept_token_count": 736.58125, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9902296423912048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009770349320024251, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.152490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.02423400878905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2500, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.09931689947843551, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.018682537600398064, + "grad_norm": 0.10374268144369125, + "kept_token_count": 737.221875, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903515577316284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009648441430181265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.95269622802735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.49480895996095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2510, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.09929787293076515, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.018592072278261186, + "grad_norm": 0.11544652283191681, + "kept_token_count": 736.928125, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9908410489559174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009158945921808482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.9130584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2247253417969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.09923015460371971, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.017570994142442942, + "grad_norm": 0.10827134549617767, + "kept_token_count": 737.16875, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9912691712379456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873081274330616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.2503662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.11586303710936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2530, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.1006508506834507, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.01756731504574418, + "grad_norm": 0.14567744731903076, + "kept_token_count": 735.7515625, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907942473888397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009205765090882778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.0026626586914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.94810180664064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2540, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10023057982325553, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.020158428605645894, + "grad_norm": 0.1141790822148323, + "kept_token_count": 735.965625, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9899950385093689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01000495497137308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.37988433837893, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0007019042969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2550, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.0998540312051773, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.01956467516720295, + "grad_norm": 0.16440068185329437, + "kept_token_count": 736.5796875, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.990196841955185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009803153574466705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.70823822021487, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2245819091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2560, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09985073953866959, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.01696142153814435, + "grad_norm": 0.1678708791732788, + "kept_token_count": 736.740625, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9910341799259186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008965817093849183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.12596588134767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7738098144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2570, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.09938783720135688, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.019487938936799766, + "grad_norm": 0.1150212362408638, + "kept_token_count": 737.190625, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.990617823600769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009382180683314801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.60023803710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1451385498047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2580, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.09991423338651657, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.0150335437618196, + "grad_norm": 0.10144449770450592, + "kept_token_count": 736.43125, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9907361149787903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009263885673135519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.43484954833986, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9987426757813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2590, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.09968986511230468, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.019641585368663072, + "grad_norm": 0.12113130837678909, + "kept_token_count": 736.746875, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914328038692475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008567184209823608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.29795227050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3167022705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2600, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.09973763450980186, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.018578110449016094, + "grad_norm": 0.11656590551137924, + "kept_token_count": 736.5984375, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9910805761814118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008919424563646316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.4978363037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.53391723632814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2610, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.10020350068807601, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.0185610082000494, + "grad_norm": 0.145968496799469, + "kept_token_count": 736.0109375, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9907793283462525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009220654517412186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.1730224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8230926513672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2620, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.10078209862113, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.017681587394326924, + "grad_norm": 0.11327216029167175, + "kept_token_count": 735.74375, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9911166310310364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0088833874091506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.3036376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5491149902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2630, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.0994064599275589, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.019192972872406244, + "grad_norm": 0.12839040160179138, + "kept_token_count": 737.240625, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9906748712062836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009325134009122849, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.74527435302736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8925262451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2640, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.100050900131464, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.01935202395543456, + "grad_norm": 0.13810528814792633, + "kept_token_count": 736.296875, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9910100519657135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008989953622221946, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.44136810302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.49268798828126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09958070665597915, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.018551858700811864, + "grad_norm": 0.12369773536920547, + "kept_token_count": 736.6890625, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9907025516033172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009297450352460147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.4215942382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7839813232422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2660, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.09996970295906067, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.018871223367750645, + "grad_norm": 0.13798530399799347, + "kept_token_count": 736.2484375, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9903373658657074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00966263934969902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.3782684326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2449127197266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2670, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.09954524710774422, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.02022105539217591, + "grad_norm": 0.1533023864030838, + "kept_token_count": 736.771875, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9906170845031739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009382921364158392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.6538330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.053466796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2680, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.1006568044424057, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.019048475567251442, + "grad_norm": 0.11896802484989166, + "kept_token_count": 736.1078125, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9910219609737396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008978024125099182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.2647994995117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.90297241210936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2690, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.10016117617487907, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.019441538490355015, + "grad_norm": 0.13291580975055695, + "kept_token_count": 736.0625, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9901166975498199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009883309155702591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.7483276367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.31121215820315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2700, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.09943175986409188, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.01749168699607253, + "grad_norm": 0.13725125789642334, + "kept_token_count": 737.0140625, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.99034463763237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00965536255389452, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.3855941772461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1383850097656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2710, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.10009532496333122, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.018285088427364826, + "grad_norm": 0.18013408780097961, + "kept_token_count": 736.196875, + "learning_rate": 2.986478491689952e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9904859900474549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009514027368277311, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.5466033935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5203063964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.09956651106476784, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.018416695669293402, + "grad_norm": 0.15299077332019806, + "kept_token_count": 736.775, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9906913757324218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00930861895903945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.72496185302737, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5958221435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2730, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09975680559873581, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.017023353185504676, + "grad_norm": 0.15002167224884033, + "kept_token_count": 736.3734375, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909430742263794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009056909661740064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.5925659179687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0860321044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2740, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.10046521052718163, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.017723689321428537, + "grad_norm": 0.1700015664100647, + "kept_token_count": 735.8859375, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9902024269104004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00979757672175765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.6569396972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2652923583984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2750, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.09993288666009903, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.01680081682279706, + "grad_norm": 0.13867102563381195, + "kept_token_count": 736.6046875, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9910292446613311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00897075729444623, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.3586364746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1869171142578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2760, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09977450966835022, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.020744918938726188, + "grad_norm": 0.14258868992328644, + "kept_token_count": 736.296875, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9908106565475464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918933404609561, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.2016235351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.42230529785155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2770, + "valid_token_count": 817.9015625 + }, + { + "actual_token_mask_ratio": 0.1002556823194027, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.017931245639920233, + "grad_norm": 0.15162575244903564, + "kept_token_count": 735.828125, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9908209621906281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009179035108536482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.0314514160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.24098205566406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2780, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.0998514361679554, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.018203623313456773, + "grad_norm": 0.13603352010250092, + "kept_token_count": 736.4328125, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9900164842605591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998351899906993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.34057769775393, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8025299072266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.09977653175592423, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.019313958846032618, + "grad_norm": 0.17452776432037354, + "kept_token_count": 736.34375, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9913984715938569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008601541910320521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.5688110351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6265472412109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2800, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.10140694677829742, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.021381164621561765, + "grad_norm": 0.13133034110069275, + "kept_token_count": 735.4734375, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9902624070644379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009737598150968552, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.5718078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7264770507812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2810, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.10019526928663254, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.019051280897110702, + "grad_norm": 0.12864120304584503, + "kept_token_count": 736.3546875, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.024, + "mgd_cos_sim": 0.990189254283905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009810759034007788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.4713928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1398986816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2820, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.09967379346489906, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.016453106701374055, + "grad_norm": 0.1186806708574295, + "kept_token_count": 736.9375, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9909359514713287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009064043313264847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.74939880371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0104553222656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2830, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.10035342499613761, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.016766002494841815, + "grad_norm": 0.10479994863271713, + "kept_token_count": 736.1921875, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9910053074359894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008994692470878363, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.5321380615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0820251464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2840, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09938364326953888, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.016807472333312035, + "grad_norm": 0.10769065469503403, + "kept_token_count": 737.00625, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9913245260715484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008675463870167733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.6154190063477, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.80247802734374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2850, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.09982041493058205, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.02086947439238429, + "grad_norm": 0.12379495799541473, + "kept_token_count": 736.80625, + "learning_rate": 2.98320044140705e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9916472673416138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00835272865369916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.55013580322264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.99383239746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2860, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.09961405619978905, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.018766173534095286, + "grad_norm": 0.12480589747428894, + "kept_token_count": 736.5625, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9912571847438812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008742818515747786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.4073516845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5971221923828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2870, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.10014883056282997, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.018001390714198352, + "grad_norm": 0.11162145435810089, + "kept_token_count": 736.1875, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.990372383594513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009627616778016091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.78114929199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5385223388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2880, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.0998690590262413, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.019897852465510367, + "grad_norm": 0.12618079781532288, + "kept_token_count": 736.546875, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9902623414993286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009737652447074651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.8296264648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.70997619628906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2890, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09964613541960717, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.01763639524579048, + "grad_norm": 0.1187954917550087, + "kept_token_count": 736.4890625, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908380389213562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009161973837763071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.1882720947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.075, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2900, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.0999560035765171, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.0169312066398561, + "grad_norm": 0.15692709386348724, + "kept_token_count": 736.69375, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9909031450748443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009096834249794484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.2619659423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7607818603516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2910, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.09998164549469948, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.021199718490242957, + "grad_norm": 0.13197295367717743, + "kept_token_count": 736.2953125, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0264, + "mgd_cos_sim": 0.989698451757431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010301548801362515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.16759338378904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0075714111328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09988167211413383, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.018838233966380357, + "grad_norm": 0.13616837561130524, + "kept_token_count": 736.859375, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9907318592071533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009268124215304851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.0388885498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2991119384766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2930, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.0996618315577507, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.021340166591107844, + "grad_norm": 0.1530594676733017, + "kept_token_count": 736.7921875, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9909654140472413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009034589398652315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.8426452636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.84119873046876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2940, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.10000793933868408, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.017563793808221817, + "grad_norm": 0.14578714966773987, + "kept_token_count": 736.709375, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907603859901428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009239629097282886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.81239166259763, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3175537109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2950, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.10001224875450135, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.01931196954101324, + "grad_norm": 0.14771795272827148, + "kept_token_count": 736.6015625, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9911105692386627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008889427501708269, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.5619262695312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7260284423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2960, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.09971096143126487, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.01753249382600188, + "grad_norm": 0.14842505753040314, + "kept_token_count": 736.684375, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9907897055149079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009210285823792219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.0213287353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2742645263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10027004852890968, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.018372665718197824, + "grad_norm": 0.13049590587615967, + "kept_token_count": 736.3546875, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9897653758525848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010234626568853855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.19685821533204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.46129150390624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.1000474140048027, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.01616821074858308, + "grad_norm": 0.1906062364578247, + "kept_token_count": 736.1671875, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9906357884407043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009364215191453695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.2543548583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.63857116699216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 2990, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.09982389658689499, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.016412969678640366, + "grad_norm": 0.12261510640382767, + "kept_token_count": 736.978125, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9906473219394684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009352684486657381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.0848083496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3802490234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3000, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.10027047023177146, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.01788167003542185, + "grad_norm": 0.19480061531066895, + "kept_token_count": 736.3296875, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.990793514251709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00920648593455553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.4304962158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4834045410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3010, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.09930597096681595, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.018929185811430216, + "grad_norm": 0.11142159253358841, + "kept_token_count": 737.45, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9892057776451111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010794212576001883, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.66802368164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.848291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3020, + "valid_token_count": 818.7578125 + }, + { + "actual_token_mask_ratio": 0.10093187615275383, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.0161429600790143, + "grad_norm": 0.12118112295866013, + "kept_token_count": 735.8875, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9904560565948486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009543950390070676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.8532653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1554931640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3030, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.1004202701151371, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.01889535952359438, + "grad_norm": 0.1195976585149765, + "kept_token_count": 736.290625, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9905315935611725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009468413051217794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.0419403076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3660491943359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3040, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.10010462254285812, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.01598623851314187, + "grad_norm": 0.12061408907175064, + "kept_token_count": 736.0703125, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9908266663551331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009173337556421756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.7832458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.395947265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3050, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.10017508938908577, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.018696373142302037, + "grad_norm": 0.13460803031921387, + "kept_token_count": 735.9796875, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9900326550006866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009967326745390892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.3506286621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7605834960938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3060, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.09992004632949829, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.017592079658061265, + "grad_norm": 0.12627162039279938, + "kept_token_count": 736.5125, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907893478870392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009210656583309173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.2445495605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.40626831054686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3070, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09981287717819214, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.01778523875400424, + "grad_norm": 0.1085885763168335, + "kept_token_count": 736.3453125, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9910927236080169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008907279651612043, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.6092041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.80205078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.10032379850745202, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.018457460682839155, + "grad_norm": 0.17800064384937286, + "kept_token_count": 735.996875, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9916952550411224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008304738719016314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.2421936035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.33014526367185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3090, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.09926663711667061, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.019583987910300492, + "grad_norm": 0.09479047358036041, + "kept_token_count": 736.8203125, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9907607316970826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009239272400736808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.26702880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5677154541016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3100, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.09961698204278946, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.01735340543091297, + "grad_norm": 0.10032481700181961, + "kept_token_count": 736.8390625, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9909672498703003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009032743144780397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.21990966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5631042480469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3110, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.10051684230566024, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.017587849963456394, + "grad_norm": 0.10573010891675949, + "kept_token_count": 736.1671875, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9911610722541809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008838918525725603, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.7110992431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3910827636719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3120, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.09989640042185784, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.020280011743307114, + "grad_norm": 0.13991305232048035, + "kept_token_count": 736.4109375, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9911164104938507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008883575443178415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.44286956787107, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.16709899902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3130, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.0997886337339878, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.0177080312743783, + "grad_norm": 0.13026845455169678, + "kept_token_count": 736.6140625, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.99113609790802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008863898552954197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.8187225341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7096862792969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3140, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10071203783154488, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.0171929438598454, + "grad_norm": 0.09541922062635422, + "kept_token_count": 735.9546875, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.991124027967453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008875966910272837, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.74122314453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0766143798828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3150, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.10008632317185402, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.017970752995461227, + "grad_norm": 0.15197359025478363, + "kept_token_count": 736.5203125, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9905386567115784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009461346827447415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.44004211425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4343566894531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3160, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.10034166499972344, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.016242476645857096, + "grad_norm": 0.09209353476762772, + "kept_token_count": 736.159375, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9910680830478669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008931921795010567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.73906555175782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4732177734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3170, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09989823848009109, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.017125525511801243, + "grad_norm": 0.11503244191408157, + "kept_token_count": 736.64375, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907243907451629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009275608789175748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.1286560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3429473876953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3180, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.09986217841506004, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.018700814340263605, + "grad_norm": 0.1286849081516266, + "kept_token_count": 736.475, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903663694858551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00963362455368042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.5262939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.65106811523435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3190, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.1004817396402359, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.017884306143969297, + "grad_norm": 0.09704916179180145, + "kept_token_count": 736.1703125, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9903949677944184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009605032112449407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.6208068847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.14317626953124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3200, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.10046142712235451, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.018003877624869345, + "grad_norm": 0.1244550347328186, + "kept_token_count": 736.0453125, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9902591347694397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009740879200398922, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.594499206543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1544586181641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09958742633461952, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.01702582947909832, + "grad_norm": 0.14032815396785736, + "kept_token_count": 736.696875, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9910309076309204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008969086315482854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.77471618652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1809326171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3220, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10021718293428421, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.014799577370285987, + "grad_norm": 0.13561765849590302, + "kept_token_count": 736.221875, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9914707899093628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008529209811240435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.70233154296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.13092346191405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3230, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.10009697154164314, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.017549639195203783, + "grad_norm": 0.13333317637443542, + "kept_token_count": 736.0390625, + "learning_rate": 2.972524844066513e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9911740362644196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008825970906764269, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.0319122314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0054656982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3240, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.09949300810694695, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.01695221783593297, + "grad_norm": 0.15518338978290558, + "kept_token_count": 736.946875, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9906414449214935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009358563646674155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.9376220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.70583190917966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3250, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.09990532696247101, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.018974354676902293, + "grad_norm": 0.13498999178409576, + "kept_token_count": 736.475, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9907907485961914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00920925410464406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.2786834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.02088623046876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3260, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.09991777464747428, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.017395757138729095, + "grad_norm": 0.10170115530490875, + "kept_token_count": 736.778125, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913978219032288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008602188341319561, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.43516845703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9258056640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3270, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.09930792972445487, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.019307829905301332, + "grad_norm": 0.12572301924228668, + "kept_token_count": 736.915625, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9902671277523041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009732888545840979, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.8672576904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7157257080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3280, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.0999229334294796, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.022704013995826243, + "grad_norm": 0.1375907063484192, + "kept_token_count": 736.5375, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0276, + "mgd_cos_sim": 0.990163904428482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009836088214069606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.3631134033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.68162231445314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3290, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.0996139444410801, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.01792035307735205, + "grad_norm": 0.14204621315002441, + "kept_token_count": 736.8859375, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9911552190780639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008844783436506987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.89986419677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9822601318359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3300, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.09979856312274933, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.01781517397612333, + "grad_norm": 0.10743262618780136, + "kept_token_count": 736.56875, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9915870249271392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008412977028638125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.18018493652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.50682067871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3310, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.09997359216213227, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.02046907562762499, + "grad_norm": 0.14949820935726166, + "kept_token_count": 736.546875, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9908530294895173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009146961383521557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.33087158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8887603759766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3320, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.09977552592754364, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.017309017945080996, + "grad_norm": 0.16084171831607819, + "kept_token_count": 736.575, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9908305406570435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009169467259198427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.4123931884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1154052734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3330, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.09980237260460853, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.01863956209272146, + "grad_norm": 0.12069318443536758, + "kept_token_count": 736.325, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9908488035202027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009151201788336038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.6131256103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.339697265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3340, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.09925105571746826, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.017638951539993286, + "grad_norm": 0.13947176933288574, + "kept_token_count": 736.853125, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.990825480222702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009174520988017321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.40675659179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.56849060058596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3350, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.10014470294117928, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.01760916840285063, + "grad_norm": 0.12560315430164337, + "kept_token_count": 736.1203125, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9906175255775451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938247973099351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.67188262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6093505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3360, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.09988282322883606, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.019070236012339592, + "grad_norm": 0.13295993208885193, + "kept_token_count": 736.4890625, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9907997131347657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009200271684676409, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.1366027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.55572204589845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3370, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.10013945996761323, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.015387126710265875, + "grad_norm": 0.11791592091321945, + "kept_token_count": 736.4421875, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9910367727279663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00896322149783373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.470280456543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6762268066406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3380, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10041306614875793, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.015964676067233084, + "grad_norm": 0.11688728630542755, + "kept_token_count": 735.9828125, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9904521107673645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009547911398112775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.0229736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7831756591797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3390, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.09990066364407539, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.01988500924780965, + "grad_norm": 0.1366073489189148, + "kept_token_count": 736.203125, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9903905749320984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009609429351985454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.2164703369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4115692138672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3400, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.1001317948102951, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.015912585239857436, + "grad_norm": 0.1387384682893753, + "kept_token_count": 736.209375, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9909802615642548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009019735921174287, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.1140502929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9494293212891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3410, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.09989911019802093, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.01539604552090168, + "grad_norm": 0.14995670318603516, + "kept_token_count": 736.5046875, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9904687881469727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009531212598085403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.9770782470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.82005615234374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3420, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.09930583238601684, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.020099888369441033, + "grad_norm": 0.13049858808517456, + "kept_token_count": 736.884375, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9904080688953399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009591937158256769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.17523498535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.40540466308596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3430, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.09986889436841011, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.021216315031051636, + "grad_norm": 0.1086823046207428, + "kept_token_count": 736.2265625, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9901823997497559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009817597921937704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.18690795898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5187683105469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3440, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.10037549287080764, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.018963928148150445, + "grad_norm": 0.13359247148036957, + "kept_token_count": 736.4375, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9903052151203156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009694783575832843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.7257537841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5574890136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3450, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.09956242814660073, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.020209626760333776, + "grad_norm": 0.13733117282390594, + "kept_token_count": 736.4421875, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9911085903644562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008891422580927611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.54978942871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8542541503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3460, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.09999757632613182, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.016491119284182787, + "grad_norm": 0.10944392532110214, + "kept_token_count": 736.603125, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.990517771244049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009482233878225088, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.8264404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7803619384766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3470, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.09974292069673538, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.01925828233361244, + "grad_norm": 0.16579149663448334, + "kept_token_count": 737.003125, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9911993086338043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008800696954131127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.6739501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1279327392578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3480, + "valid_token_count": 818.6578125 + }, + { + "actual_token_mask_ratio": 0.09941122308373451, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.01698103155940771, + "grad_norm": 0.14656302332878113, + "kept_token_count": 736.94375, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9907779216766357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0092220907099545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.5788543701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.81244201660155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3490, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.10018776804208755, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.01732523860409856, + "grad_norm": 0.14771591126918793, + "kept_token_count": 736.375, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9907545566558837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009245443437248469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.73524475097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.83912048339846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3500, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.0999699242413044, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.016830039490014315, + "grad_norm": 0.11180265247821808, + "kept_token_count": 736.3953125, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9909874975681305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009012508299201727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.0446838378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3037139892578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3510, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10052573084831237, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.01981815751641989, + "grad_norm": 0.14744441211223602, + "kept_token_count": 736.078125, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9913858354091645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008614172227680683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.5900573730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.81543884277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3520, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.10041772052645684, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.019437560997903346, + "grad_norm": 0.1268990933895111, + "kept_token_count": 736.2109375, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9913335263729095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008666457422077656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.77294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9139038085938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.09987665638327599, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.017493678070604802, + "grad_norm": 0.12042878568172455, + "kept_token_count": 736.9578125, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9908333480358124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009166652522981166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.59568786621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9065887451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3540, + "valid_token_count": 818.73125 + }, + { + "actual_token_mask_ratio": 0.09974445775151253, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.017736657056957483, + "grad_norm": 0.11705845594406128, + "kept_token_count": 736.63125, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.990103018283844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009896981436759234, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.6595886230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5326416015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3550, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10051507130265236, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.018259658571332694, + "grad_norm": 0.1708606332540512, + "kept_token_count": 735.91875, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.023, + "mgd_cos_sim": 0.990579628944397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009420363884419202, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.90403442382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8417510986328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3560, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.10003653764724732, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.018316812068223953, + "grad_norm": 0.11453288793563843, + "kept_token_count": 736.3453125, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9909530639648437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00904695326462388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.6290557861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.99167785644534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.09992212206125259, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.019874213635921477, + "grad_norm": 0.11322524398565292, + "kept_token_count": 736.7375, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9904746413230896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009525362774729729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.0339874267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9478485107422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3580, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.09960659816861153, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.01779921669512987, + "grad_norm": 0.1161990687251091, + "kept_token_count": 736.865625, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9906729221343994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009327081870287657, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.2861053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9418487548828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3590, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.10018117427825927, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.019548140186816453, + "grad_norm": 0.1314779371023178, + "kept_token_count": 736.328125, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9905403971672058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009459605067968368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.34954223632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5055297851562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3600, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09977272599935531, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.01703464603051543, + "grad_norm": 0.127919003367424, + "kept_token_count": 736.396875, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909432291984558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009056778997182846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.55196380615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.85048522949216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3610, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.1006766565144062, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.015558455791324376, + "grad_norm": 0.13665199279785156, + "kept_token_count": 735.4953125, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9911767601966858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008823247440159322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.25635986328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.10132751464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3620, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.1000713437795639, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.01680281711742282, + "grad_norm": 0.14750869572162628, + "kept_token_count": 736.2109375, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9910919070243835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008908107876777649, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.70946044921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8895965576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3630, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.09988771826028824, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.018008433934301137, + "grad_norm": 0.1303689330816269, + "kept_token_count": 736.80625, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9904350280761719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009564981330186129, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.8603210449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3056182861328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3640, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.10036093592643738, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.019673653692007066, + "grad_norm": 0.19166865944862366, + "kept_token_count": 736.06875, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9906701624393464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009329827036708593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.4094970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.39434204101565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3650, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.09996303915977478, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.018693684507161377, + "grad_norm": 0.16338728368282318, + "kept_token_count": 736.48125, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9907822787761689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00921771042048931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.3254150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0166259765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.1002474769949913, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.018879633396863937, + "grad_norm": 0.13995617628097534, + "kept_token_count": 736.265625, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9908126175403595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009187368117272854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.61842956542966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.83005676269534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3670, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.10000778660178185, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.016718839202076195, + "grad_norm": 0.14958499372005463, + "kept_token_count": 736.7671875, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.990016770362854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009983224514871835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.3811401367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.85853271484376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3680, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.10048330500721932, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.01730442326515913, + "grad_norm": 0.11458824574947357, + "kept_token_count": 736.034375, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9907963573932648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009203641209751368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.9200103759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7941833496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3690, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10015752017498017, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.019053922779858113, + "grad_norm": 0.10131317377090454, + "kept_token_count": 736.6234375, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9909439444541931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00905608180910349, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.01613464355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.59984130859374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3700, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.09976509138941765, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.018411446642130614, + "grad_norm": 0.11289618164300919, + "kept_token_count": 736.5515625, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9902856767177581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00971432039514184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.9001892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.46660766601565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3710, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10003534853458404, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.019707252364605667, + "grad_norm": 0.11448097974061966, + "kept_token_count": 736.53125, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9902554810047149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0097445173189044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.8686126708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36122741699216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3720, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.10039677992463111, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.01797666968777776, + "grad_norm": 0.11452915519475937, + "kept_token_count": 736.1046875, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9906768083572388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009323190804570914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.8746032714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.72801513671874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3730, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09963932558894158, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.01758129009976983, + "grad_norm": 0.10750444233417511, + "kept_token_count": 736.79375, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.990574711561203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00942528136074543, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.4939208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3945770263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3740, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.10016562268137932, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.019308339338749647, + "grad_norm": 0.12916584312915802, + "kept_token_count": 735.9234375, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9909633278846741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00903667053207755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.3130737304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7673400878906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3750, + "valid_token_count": 817.840625 + }, + { + "actual_token_mask_ratio": 0.10045975297689438, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.018307531625032424, + "grad_norm": 0.1403598189353943, + "kept_token_count": 736.0640625, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9906269788742066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00937302215024829, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.4912170410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.34702758789064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3760, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.10031156539916992, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.017043178621679544, + "grad_norm": 0.1038823053240776, + "kept_token_count": 735.9671875, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9912103414535522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008789658453315497, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.37046203613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.947998046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3770, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.1001846507191658, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.016465229727327824, + "grad_norm": 0.12625831365585327, + "kept_token_count": 736.421875, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9907905995845795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009209388121962547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.40366821289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58807678222655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3780, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.09908592477440833, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.018423198908567428, + "grad_norm": 0.13417163491249084, + "kept_token_count": 737.0640625, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9910851418972015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008914853818714619, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.8042327880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.03997192382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3790, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.09984997734427452, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.018343734182417394, + "grad_norm": 0.15946078300476074, + "kept_token_count": 736.4703125, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.990876990556717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00912299994379282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.4548095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7399444580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3800, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.10007108971476555, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.01651483038440347, + "grad_norm": 0.12186123430728912, + "kept_token_count": 736.1578125, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9907469093799591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009253090899437665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.11143493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.23597106933596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3810, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.09966341480612755, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.01821138383820653, + "grad_norm": 0.11743904650211334, + "kept_token_count": 736.71875, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9911767661571502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008823230490088462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.04198303222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1338623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3820, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10030103325843812, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.01825149478390813, + "grad_norm": 0.10920114070177078, + "kept_token_count": 736.35625, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9914829432964325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008517046831548214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.79334411621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3996154785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3830, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.1004727378487587, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.017440368235111237, + "grad_norm": 0.11074068397283554, + "kept_token_count": 736.0109375, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9908643186092376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009135682694613933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.6657348632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.32456970214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3840, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.10044616982340812, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.01764281541109085, + "grad_norm": 0.12459435313940048, + "kept_token_count": 736.1515625, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908102691173554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918973032385111, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.4027069091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.78358154296876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3850, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.10049774944782257, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.018129661958664656, + "grad_norm": 0.1320890486240387, + "kept_token_count": 735.790625, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.023, + "mgd_cos_sim": 0.99034885764122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00965113528072834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.14874267578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.07505798339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3860, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.0996415451169014, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.01997614149004221, + "grad_norm": 0.12647981941699982, + "kept_token_count": 736.728125, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9906498670578003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00935013424605131, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.92548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8506683349609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3870, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.09958532527089119, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.018005120009183882, + "grad_norm": 0.11653801053762436, + "kept_token_count": 736.71875, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9913345038890838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008665497321635485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.7392974853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1149017333984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3880, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.10047695562243461, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.017012884933501483, + "grad_norm": 0.13757991790771484, + "kept_token_count": 736.0546875, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9911164224147797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008883574791252614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.1146575927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05016174316404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3890, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10035628825426102, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.0188037920743227, + "grad_norm": 0.14463819563388824, + "kept_token_count": 736.31875, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.989968341588974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010031668655574321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.7354217529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.81044311523436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3900, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.10010468661785125, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.020843626372516154, + "grad_norm": 0.1253921389579773, + "kept_token_count": 736.4765625, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.991067773103714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00893223499879241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.5723541259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3833251953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3910, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.0999882087111473, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.019631539657711982, + "grad_norm": 0.139969602227211, + "kept_token_count": 736.4453125, + "learning_rate": 2.946988727146143e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9913365364074707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00866346452385187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.82762145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5685638427734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3920, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.09969363436102867, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.01624243436381221, + "grad_norm": 0.13473668694496155, + "kept_token_count": 736.2328125, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9918235242366791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008176473248749972, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.78339233398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9307922363281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3930, + "valid_token_count": 817.759375 + }, + { + "actual_token_mask_ratio": 0.10065561532974243, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.0170830219052732, + "grad_norm": 0.1611064076423645, + "kept_token_count": 735.796875, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9906369090080261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009363087359815836, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.2269226074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5138397216797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10045786798000336, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.017185136396437883, + "grad_norm": 0.1281617134809494, + "kept_token_count": 736.0140625, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907386779785157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009261319041252136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.19219970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.469189453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3950, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.09941739141941071, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.01935352720320225, + "grad_norm": 0.14059685170650482, + "kept_token_count": 736.809375, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9901781916618347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00982181318104267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.0333312988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1737457275391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3960, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.10009296238422394, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.018074749130755664, + "grad_norm": 0.13789182901382446, + "kept_token_count": 736.371875, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.990869814157486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009130180627107621, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.4311859130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.88294677734376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09945437163114548, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.02383789885789156, + "grad_norm": 0.13917924463748932, + "kept_token_count": 737.0125, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0284, + "mgd_cos_sim": 0.9909228026866913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009077189210802316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.90821838378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2230621337891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3980, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.10033580958843231, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.021450463309884073, + "grad_norm": 0.14790506660938263, + "kept_token_count": 736.0859375, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9908324241638183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009167592320591212, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.1377349853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.16600341796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 3990, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.09936547204852104, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.018645609728991985, + "grad_norm": 0.09085589647293091, + "kept_token_count": 736.60625, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.023, + "mgd_cos_sim": 0.991322147846222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008677843026816845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.5448913574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.27744140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4000, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.10051095485687256, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.01967667480930686, + "grad_norm": 0.12645772099494934, + "kept_token_count": 735.8984375, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.99106764793396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0089323528110981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.1786834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2755523681641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4010, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09997195675969124, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.017064095195382833, + "grad_norm": 0.12700259685516357, + "kept_token_count": 736.728125, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.991609251499176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008390746731311082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.9822631835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1283020019531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4020, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.10023821741342545, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.01615453455597162, + "grad_norm": 0.12152964621782303, + "kept_token_count": 736.4390625, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9910402417182922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008959743846207856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.16033325195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0208374023438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4030, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.0996767096221447, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.017575164139270783, + "grad_norm": 0.11882390826940536, + "kept_token_count": 736.475, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9906680524349213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009331961628049612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.01218566894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.86533203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4040, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.10036000609397888, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.015989204868674278, + "grad_norm": 0.11341783404350281, + "kept_token_count": 736.2796875, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9912775993347168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008722401689738035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.58766784667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.63433837890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.09998553916811943, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.01751160016283393, + "grad_norm": 0.11382917314767838, + "kept_token_count": 736.2890625, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9911612570285797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008838744089007377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.46954956054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.364599609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4060, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09983134120702744, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.018210010696202518, + "grad_norm": 0.12807051837444305, + "kept_token_count": 736.5671875, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9907924234867096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009207584895193577, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.97825012207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9270080566406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4070, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.09991731271147727, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.017775773908942936, + "grad_norm": 0.10437645018100739, + "kept_token_count": 736.7015625, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9909356653690338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009064347483217716, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.6765075683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8788360595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4080, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.09931840002536774, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.017430211883038282, + "grad_norm": 0.17382217943668365, + "kept_token_count": 736.6328125, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9907478511333465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009252151567488908, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.9165100097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6791717529297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4090, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.10082717016339302, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.018638925068080424, + "grad_norm": 0.11431209743022919, + "kept_token_count": 735.696875, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9906150102615356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009384996257722378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.3393035888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6639862060547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4100, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.10036494582891464, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.018112821970134974, + "grad_norm": 0.10850103199481964, + "kept_token_count": 736.5234375, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906067550182343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009393235668540001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.17713317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.72617492675784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4110, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.09993026107549667, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.017457062471657993, + "grad_norm": 0.12401090562343597, + "kept_token_count": 736.41875, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9909997880458832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900023216381669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.95709838867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.44676513671874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4120, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.09939049780368805, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.015542091336101294, + "grad_norm": 0.11247029900550842, + "kept_token_count": 736.88125, + "learning_rate": 2.937452366010036e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9910076558589935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00899234488606453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.95774536132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6567626953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4130, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.10045718848705291, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.021496732905507087, + "grad_norm": 0.10891535878181458, + "kept_token_count": 736.159375, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9911488056182861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008851198572665453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.46592407226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8068450927734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4140, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.09979586228728295, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.017686854861676694, + "grad_norm": 0.14288516342639923, + "kept_token_count": 736.453125, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9902069687843322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009793033078312873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.6267822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8843231201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4150, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.09943647235631943, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.020138866920024156, + "grad_norm": 0.12122058123350143, + "kept_token_count": 736.8484375, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9904361248016358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009563858062028885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.99318542480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.57435913085936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4160, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.0995872862637043, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.01864970875903964, + "grad_norm": 0.11109218001365662, + "kept_token_count": 736.54375, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9907647967338562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009235202893614769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.4651245117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3447326660156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4170, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.10006930753588676, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.017887497693300246, + "grad_norm": 0.12687161564826965, + "kept_token_count": 736.3421875, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9908786714076996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009121331945061683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.4388793945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.28196411132814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4180, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.10054124891757965, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.01672378657385707, + "grad_norm": 0.10839472711086273, + "kept_token_count": 736.2640625, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9905293107032775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009470704104751348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.5563171386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1074157714844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4190, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.09975863695144653, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.016502228658646347, + "grad_norm": 0.15627284348011017, + "kept_token_count": 736.5328125, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9905641973018646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009435806423425674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.66722717285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3308502197266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4200, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10003243237733841, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.016689721029251814, + "grad_norm": 0.11838741600513458, + "kept_token_count": 736.3453125, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9907769978046417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00922300824895501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.6571533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6077606201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4210, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.1009541317820549, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.016464110650122165, + "grad_norm": 0.14183054864406586, + "kept_token_count": 735.8515625, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9901198029518128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009880193416029215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.8825286865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0545379638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4220, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.09991946220397949, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.016803459636867045, + "grad_norm": 0.1450086385011673, + "kept_token_count": 736.8515625, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9904182255268097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009581770375370979, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.4700500488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.25831909179686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4230, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.09980940446257591, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.01925210664048791, + "grad_norm": 0.14132556319236755, + "kept_token_count": 736.665625, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9913289368152618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008671052381396293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.7736114501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6169006347656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4240, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.10051367208361625, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.016974957287311555, + "grad_norm": 0.11919330060482025, + "kept_token_count": 736.034375, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9911900758743286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00880993315950036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.120703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9408874511719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4250, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.10046601593494416, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.01662121042609215, + "grad_norm": 0.11591967940330505, + "kept_token_count": 736.071875, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.021, + "mgd_cos_sim": 0.991240406036377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008759604766964912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.31971740722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10943298339845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4260, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.10030615627765656, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.014559788443148137, + "grad_norm": 0.11859825998544693, + "kept_token_count": 736.1203125, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9909624516963959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009037544298917056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.7430053710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9052307128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4270, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.1000811755657196, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.018471291195601224, + "grad_norm": 0.16676075756549835, + "kept_token_count": 736.434375, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9905472338199616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009452773351222276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.3062286376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5641723632813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4280, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1002542681992054, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.01878976756706834, + "grad_norm": 0.12217406183481216, + "kept_token_count": 736.321875, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9906373441219329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009362651966512203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.4832794189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1362640380859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4290, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.09964662864804268, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.01825620364397764, + "grad_norm": 0.1198226660490036, + "kept_token_count": 736.6109375, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9906554043292999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009344599302858114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.09047241210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.53358764648436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4300, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.0993241861462593, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.01848001927137375, + "grad_norm": 0.14259473979473114, + "kept_token_count": 737.2171875, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9896088302135467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010391172487288713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.0393096923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8482604980469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4310, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.1006189726293087, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.016135644167661667, + "grad_norm": 0.0916687622666359, + "kept_token_count": 735.840625, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9909373342990875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009062659461051226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.9540557861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.11163940429685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4320, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.10035477355122566, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.018457674700766803, + "grad_norm": 0.13929060101509094, + "kept_token_count": 736.1140625, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9907164752483368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009283526428043842, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.54644470214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.60564575195315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4330, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.10018882751464844, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.019245321862399578, + "grad_norm": 0.14333006739616394, + "kept_token_count": 735.89375, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9903877913951874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009612192958593368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.05230102539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9415679931641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4340, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.10066860318183898, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.017424239404499532, + "grad_norm": 0.1077875867486, + "kept_token_count": 735.828125, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9906570136547088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009342982154339551, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.2028747558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1020263671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09997950047254563, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.01725341221317649, + "grad_norm": 0.13569582998752594, + "kept_token_count": 736.2703125, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.990861052274704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009138962253928185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.7688232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.75425720214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4360, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.10003365501761437, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.017991018202155828, + "grad_norm": 0.14442937076091766, + "kept_token_count": 736.4625, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9913612604141235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008638739492744208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.2911376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2485321044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4370, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.10018670186400414, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.019504979718476534, + "grad_norm": 0.13998214900493622, + "kept_token_count": 736.2453125, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9899397313594818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010060265380889177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.6128173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.60903015136716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4380, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.10043907389044762, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.018490356300026177, + "grad_norm": 0.1405317634344101, + "kept_token_count": 736.29375, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9908661425113678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009133865125477314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.4036376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8596984863281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4390, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.09955674931406974, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.019302584417164325, + "grad_norm": 0.13899625837802887, + "kept_token_count": 737.0734375, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9905017793178559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009498224314302205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.2766418457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.86863708496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4400, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.09991404935717582, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.020225731749087573, + "grad_norm": 0.1631270796060562, + "kept_token_count": 736.79375, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9907920360565186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009207963757216931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.2141082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5052062988281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.10060318037867547, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.018986174277961253, + "grad_norm": 0.11492699384689331, + "kept_token_count": 736.2953125, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.990655642747879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009344372991472482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.2702301025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0732818603516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4420, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.09974076449871064, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.01886403476819396, + "grad_norm": 0.16290365159511566, + "kept_token_count": 736.6515625, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9905079483985901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009492044895887375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.7427154541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.77442626953126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4430, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10027601420879365, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.018332364968955517, + "grad_norm": 0.12518703937530518, + "kept_token_count": 736.471875, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9903043985366822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009695619810372591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.7866607666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2520782470703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4440, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.09973805099725723, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.017587402369827032, + "grad_norm": 0.1330440491437912, + "kept_token_count": 736.61875, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907388806343078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00926111601293087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.50440368652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0004943847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.10007417947053909, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.01601498955860734, + "grad_norm": 0.11314190179109573, + "kept_token_count": 735.975, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910732567310333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008926740568131208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.38389587402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.385595703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4460, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.10001238137483597, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.016346720233559608, + "grad_norm": 0.10826753824949265, + "kept_token_count": 736.375, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.990979140996933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00902085890993476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.74229125976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.91420288085936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4470, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.0998660683631897, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.018938051536679268, + "grad_norm": 0.16749323904514313, + "kept_token_count": 736.5359375, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9906178534030914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009382146690040827, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.8939697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36761779785155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4480, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09984968528151512, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.018356748018413784, + "grad_norm": 0.1545967161655426, + "kept_token_count": 736.3609375, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9907224118709564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00927758552134037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.8100555419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.69737243652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4490, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.09993074238300323, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.0194969542324543, + "grad_norm": 0.15754427015781403, + "kept_token_count": 736.35, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9909796237945556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009020366333425045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.27866516113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4698089599609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4500, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10010464116930962, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.015048954728990793, + "grad_norm": 0.15060560405254364, + "kept_token_count": 736.4484375, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9903355538845062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009664447233080863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.4966522216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.71009521484376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4510, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.09964254572987556, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.016704384703189135, + "grad_norm": 0.12508030235767365, + "kept_token_count": 736.6453125, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9903369784355164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009663011506199836, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.7393402099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.50490112304686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4520, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09974074438214302, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.0166363675147295, + "grad_norm": 0.11992136389017105, + "kept_token_count": 736.4265625, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9903598248958587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009640192333608865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.5138916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7019409179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.0991699032485485, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.017984881531447172, + "grad_norm": 0.13667963445186615, + "kept_token_count": 737.0640625, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9903120577335358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009687947295606136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.70672912597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.33250427246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4540, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.10091184377670288, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.018289059028029443, + "grad_norm": 0.13789117336273193, + "kept_token_count": 735.7171875, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9903546690940856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009645336773246527, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.73297119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0631103515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4550, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10044528096914292, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.018462246656417845, + "grad_norm": 0.13954365253448486, + "kept_token_count": 735.9578125, + "learning_rate": 2.915528284197554e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9908826172351837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00911739096045494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.2029083251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5753570556641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4560, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.09962773472070693, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.016068897023797036, + "grad_norm": 0.11824842542409897, + "kept_token_count": 736.7203125, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9907443463802338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009255654644221067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.9709106445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.673095703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4570, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.09970997050404548, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.019735913071781398, + "grad_norm": 0.11633726954460144, + "kept_token_count": 736.384375, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9908734321594238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009126564394682646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.7668090820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9097595214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4580, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.09995941445231438, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.020549089182168245, + "grad_norm": 0.12847378849983215, + "kept_token_count": 736.584375, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.990396112203598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009603892918676138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.8112335205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.67271118164064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4590, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.10009500682353974, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.016470124293118716, + "grad_norm": 0.132480189204216, + "kept_token_count": 736.11875, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9911951124668121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008804875053465366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.36126098632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3357360839844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4600, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.09953858852386474, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.019665827602148057, + "grad_norm": 0.1283418983221054, + "kept_token_count": 736.79375, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.990687084197998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009312907047569752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.9599212646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7302947998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4610, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.09996388554573059, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.017390211764723062, + "grad_norm": 0.14882469177246094, + "kept_token_count": 736.290625, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9909294843673706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009070510510355234, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.76609191894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7116424560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4620, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.10038907751441002, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.01728083435446024, + "grad_norm": 0.12028706073760986, + "kept_token_count": 736.23125, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9910775184631347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008922482281923294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.3886749267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9294738769531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4630, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.09968859255313874, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.01600876385346055, + "grad_norm": 0.11241117119789124, + "kept_token_count": 736.7765625, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910719752311706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008928021974861622, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.4990570068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6087249755859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4640, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.10008048266172409, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.017587594874203204, + "grad_norm": 0.12960492074489594, + "kept_token_count": 736.275, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907071948051452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009292811807245015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.2935546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.24155578613284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09982787147164345, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.01685236729681492, + "grad_norm": 0.159835547208786, + "kept_token_count": 736.7921875, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9913844525814056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00861554555594921, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.03985595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.53961791992185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4660, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.09971433356404305, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.018453404866158964, + "grad_norm": 0.12680165469646454, + "kept_token_count": 736.6078125, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9918651700019836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00813481593504548, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.6226776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7610107421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4670, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10013686120510101, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.017373899463564157, + "grad_norm": 0.15594999492168427, + "kept_token_count": 736.4875, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9911349892616272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00886500347405672, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.47962646484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.44111328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4680, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.10031530559062958, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.019400098267942666, + "grad_norm": 0.14599750936031342, + "kept_token_count": 736.23125, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9910819947719574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008918001875281334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.7688323974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.40183715820314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4690, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10001996755599976, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.019664201233536006, + "grad_norm": 0.11280055344104767, + "kept_token_count": 736.275, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9902888178825379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009711184818297625, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.4089080810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2940338134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4700, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10012430027127266, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.01825409149751067, + "grad_norm": 0.1420060694217682, + "kept_token_count": 736.384375, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9902037620544434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009796229097992182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.5787811279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5071044921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4710, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09985871613025665, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.01877536727115512, + "grad_norm": 0.1630275696516037, + "kept_token_count": 736.9890625, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9900565803050995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009943417832255363, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.46014099121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.23900451660154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4720, + "valid_token_count": 818.75 + }, + { + "actual_token_mask_ratio": 0.09994857534766197, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.01805587699636817, + "grad_norm": 0.142839714884758, + "kept_token_count": 736.26875, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915334343910217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008466564770787954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.2857727050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.77650146484376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4730, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.10024517327547074, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.01816965052857995, + "grad_norm": 0.10031630843877792, + "kept_token_count": 736.209375, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.990447723865509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009552273713052273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.98073120117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3967529296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4740, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09969091042876244, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.020657223463058472, + "grad_norm": 0.16929654777050018, + "kept_token_count": 736.8921875, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9898288428783417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01017115442082286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.9630126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0498291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4750, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.10039490982890129, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.01900557745248079, + "grad_norm": 0.12658770382404327, + "kept_token_count": 735.9625, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9907077372074127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009292256087064743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.31414184570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.409228515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.09939023554325103, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.01841755574569106, + "grad_norm": 0.13314367830753326, + "kept_token_count": 736.7765625, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9904456019401551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009554380550980568, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.17201538085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.14869995117186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4770, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.099910469353199, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.020478235371410847, + "grad_norm": 0.11513199657201767, + "kept_token_count": 736.5859375, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9905224859714508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009477523248642683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.35516967773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.28309020996096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4780, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.10062971785664558, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.019700164347887038, + "grad_norm": 0.12636302411556244, + "kept_token_count": 736.0171875, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9899601638317108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010039842315018177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.02503356933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7333526611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4790, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.10030668750405311, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.016023567412048578, + "grad_norm": 0.13860590755939484, + "kept_token_count": 736.5125, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9912925124168396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008707490097731353, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.8531524658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2426788330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4800, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.09963157102465629, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.020721126068383455, + "grad_norm": 0.1613806039094925, + "kept_token_count": 736.875, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9908454239368438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009154584817588329, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.8254333496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.532763671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4810, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.1001588024199009, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.018478766921907662, + "grad_norm": 0.14414972066879272, + "kept_token_count": 736.3703125, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912029325962066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008797084167599677, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.81049194335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.95054626464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4820, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.10021793022751808, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.018826514948159456, + "grad_norm": 0.12726841866970062, + "kept_token_count": 736.446875, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9907436847686768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009256328269839287, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.24171752929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.83507080078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4830, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.10035164952278137, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.018914510402828454, + "grad_norm": 0.11280634999275208, + "kept_token_count": 736.1796875, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9909706473350525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009029356669634581, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.6143341064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0217529296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4840, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09968424439430237, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.017048169393092394, + "grad_norm": 0.11113926023244858, + "kept_token_count": 736.81875, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9911075115203858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00889247953891754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.2830871582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0130676269531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4850, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09980488196015358, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.019229695852845907, + "grad_norm": 0.12048310041427612, + "kept_token_count": 736.715625, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9906942784786225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00930571760982275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.2752349853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4314758300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4860, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.10011076778173447, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.01893938547000289, + "grad_norm": 0.1019439846277237, + "kept_token_count": 736.3765625, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9909443020820617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009055703785270453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.5928466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.75323791503905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4870, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.09981646165251731, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.017833512648940085, + "grad_norm": 0.11792341619729996, + "kept_token_count": 736.4515625, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9905593514442443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009440646227449178, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.15575256347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.87086791992186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4880, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.09935291782021523, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.017020328901708126, + "grad_norm": 0.10040321201086044, + "kept_token_count": 736.86875, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909073233604431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009092675894498825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.86175231933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6485870361328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4890, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1003602497279644, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.017231762316077946, + "grad_norm": 0.10506404936313629, + "kept_token_count": 736.265625, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9912123322486878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008787661325186491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.5224853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2332305908203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4900, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10006965696811676, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.01652425779029727, + "grad_norm": 0.12471218407154083, + "kept_token_count": 736.165625, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9912246525287628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008775359205901622, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.7399658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2655731201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4910, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.1001130647957325, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.021569079626351594, + "grad_norm": 0.12030215561389923, + "kept_token_count": 736.09375, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9901932835578918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009806721564382314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.9799468994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.85048217773436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4920, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.10034835413098335, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.019697361905127764, + "grad_norm": 0.12796029448509216, + "kept_token_count": 736.21875, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9915685176849365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008431492745876313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.3215301513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5945068359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4930, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.10048765242099762, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.01862852843478322, + "grad_norm": 0.16592267155647278, + "kept_token_count": 736.2078125, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903493642807006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009650622587651014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.549609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0091064453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4940, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.10021709874272347, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.017840074095875024, + "grad_norm": 0.11049992591142654, + "kept_token_count": 736.2359375, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913244187831879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008675589598715306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.87009887695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3144134521484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.0999078944325447, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.017010425869375467, + "grad_norm": 0.11883632838726044, + "kept_token_count": 736.7203125, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9909314632415771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00906852250918746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.08031311035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8507537841797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4960, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.09941569417715072, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.017988656368106603, + "grad_norm": 0.11805172264575958, + "kept_token_count": 736.8984375, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9910878777503968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008912133146077394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.76661071777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3582763671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4970, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.09992243200540543, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.018385173473507165, + "grad_norm": 0.11901780962944031, + "kept_token_count": 736.4015625, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9913607895374298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008639216050505638, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.38312072753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4015655517578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.1006238617002964, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.01834840727970004, + "grad_norm": 0.13360430300235748, + "kept_token_count": 735.8484375, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9912100911140442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008789908885955811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.5099182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9367370605469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 4990, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10032464563846588, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.018027001712471246, + "grad_norm": 0.12414922565221786, + "kept_token_count": 736.4375, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9914530277252197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008546966034919024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.48744812011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5499694824219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5000, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.10023159757256508, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.019205387495458127, + "grad_norm": 0.1196700781583786, + "kept_token_count": 736.3984375, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9915395975112915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008460399508476258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.1211212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1919891357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5010, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.09911451563239097, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.01939512984827161, + "grad_norm": 0.17389141023159027, + "kept_token_count": 737.1515625, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9909980535507202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009001938719302416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.05405883789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1702819824219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10067053586244583, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.018421673960983754, + "grad_norm": 0.14316695928573608, + "kept_token_count": 736.1625, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9913297474384308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008670251909643411, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.27635192871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5762145996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5030, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1007768154144287, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.01748016504570842, + "grad_norm": 0.10473109781742096, + "kept_token_count": 735.978125, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9913877010345459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008612290862947703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.3090423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.51654663085935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5040, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.10021337568759918, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.016873227432370186, + "grad_norm": 0.13097676634788513, + "kept_token_count": 736.5671875, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9912368416786194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00876315524801612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.07582092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4247253417969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5050, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.09993997141718865, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.01664596050977707, + "grad_norm": 0.16274596750736237, + "kept_token_count": 736.3125, + "learning_rate": 2.886040660049283e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912054538726807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008794553019106388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.11053161621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8738098144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5060, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.10014375373721122, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.016604536585509776, + "grad_norm": 0.12049054354429245, + "kept_token_count": 736.525, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9910755515098572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008924439828842879, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.29042053222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.66270751953124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5070, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.10027979239821434, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.01715701585635543, + "grad_norm": 0.11901043355464935, + "kept_token_count": 736.3921875, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9908949255943298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0091050760820508, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.4138458251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8212646484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.10011969208717346, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.018079695012420415, + "grad_norm": 0.128876194357872, + "kept_token_count": 736.2796875, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9913596630096435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00864034853875637, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.1482452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7090240478516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5090, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09941518530249596, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.017872885800898075, + "grad_norm": 0.12073457986116409, + "kept_token_count": 736.6875, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9907800257205963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0092199744656682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.1969390869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9663879394531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.1008247584104538, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.019663873687386514, + "grad_norm": 0.1402571052312851, + "kept_token_count": 735.7703125, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9903888583183289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009611133579164743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.85390014648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.51263122558595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09978379681706429, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.01768897632136941, + "grad_norm": 0.1395234316587448, + "kept_token_count": 736.5265625, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9904999732971191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009500026889145374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.74684143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9082305908203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5120, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09941472113132477, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.018159406166523694, + "grad_norm": 0.12980321049690247, + "kept_token_count": 737.1046875, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9912258744239807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008774134423583746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.5910675048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.01463623046874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5130, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.09964021220803261, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.021456471737474204, + "grad_norm": 0.13556455075740814, + "kept_token_count": 736.7796875, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9907078444957733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009292169567197561, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.42218017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4254547119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5140, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09881713762879371, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.019243918359279633, + "grad_norm": 0.11253488063812256, + "kept_token_count": 737.1328125, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9908159077167511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918407430872321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.0834655761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.82555236816404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5150, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.10007680505514145, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.019170101545751096, + "grad_norm": 0.12033671885728836, + "kept_token_count": 736.6203125, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.990528118610382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00947188064455986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6095367431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1642639160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5160, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.09925291314721107, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.017355497647076844, + "grad_norm": 0.1305002123117447, + "kept_token_count": 737.0640625, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9914746582508087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008525321073830128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.777587890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.84630126953124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5170, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.10020820647478104, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.015722785517573355, + "grad_norm": 0.07795266807079315, + "kept_token_count": 736.2703125, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9907878577709198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009212120622396468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.1318756103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.06061401367185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5180, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.09998713210225105, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.019640029594302177, + "grad_norm": 0.11398065835237503, + "kept_token_count": 736.49375, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9907802164554596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00921978782862425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.4894561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8707641601562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5190, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.10018453672528267, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.016688521578907968, + "grad_norm": 0.10353704541921616, + "kept_token_count": 736.3484375, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9909127235412598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009087290056049824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.00452880859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3461181640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5200, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09910910204052925, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.017185684014111757, + "grad_norm": 0.13575594127178192, + "kept_token_count": 737.0109375, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9913502693176269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008649727609008551, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.7213195800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1830596923828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5210, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.09970924928784371, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.017351914662867785, + "grad_norm": 0.13915443420410156, + "kept_token_count": 736.70625, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913617253303528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008638282120227814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.3747497558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.01905822753906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5220, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10024035349488258, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.018270785734057427, + "grad_norm": 0.13704705238342285, + "kept_token_count": 736.1765625, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9909510493278504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009048945363610983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.5603363037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9973571777344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5230, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10019247084856034, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.016651641111820936, + "grad_norm": 0.11379135400056839, + "kept_token_count": 736.2796875, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9908094644546509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009190538804978133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.35125122070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.550927734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5240, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10007985904812813, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.01770151751115918, + "grad_norm": 0.09521355479955673, + "kept_token_count": 736.715625, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9915185928344726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008481391705572605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.20721740722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2635528564453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5250, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.09849005565047264, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.020276350434869528, + "grad_norm": 0.1650000512599945, + "kept_token_count": 737.746875, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.990909218788147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009090776834636927, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.4964111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0421173095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5260, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.10041719749569893, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.0176972521468997, + "grad_norm": 0.133853480219841, + "kept_token_count": 736.33125, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9912937223911286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008706291019916535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.2534637451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.05311279296876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5270, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.09981721565127373, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.017607168201357125, + "grad_norm": 0.1252264827489853, + "kept_token_count": 736.3265625, + "learning_rate": 2.871726819936184e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9912915587425232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008708436321467162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.8358917236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7131072998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5280, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.1004636749625206, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.01639703568071127, + "grad_norm": 0.12051668763160706, + "kept_token_count": 736.0, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9908333241939544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009166688192635775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.92181701660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.97470703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5290, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.10007255896925926, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.017238774523139, + "grad_norm": 0.11073440313339233, + "kept_token_count": 736.6265625, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9908956170082093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009104384016245604, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.9985687255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8041046142578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5300, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.0999907061457634, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.01667987834662199, + "grad_norm": 0.0948675274848938, + "kept_token_count": 736.31875, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.021, + "mgd_cos_sim": 0.991353440284729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008646555058658124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.05409851074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3590362548828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5310, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.09910916462540627, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.018491934332996608, + "grad_norm": 0.1228189468383789, + "kept_token_count": 737.43125, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.990794014930725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009205992426723241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.7998779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.47092895507814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5320, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.09922190681099892, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.016258068662136794, + "grad_norm": 0.11755695194005966, + "kept_token_count": 737.25625, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9913065791130066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008693425077944994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.0805236816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4897125244141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.09964766204357148, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.01661886116489768, + "grad_norm": 0.15577930212020874, + "kept_token_count": 736.71875, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9911933541297913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008806661982089282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.42474975585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5191253662109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5340, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10040790513157845, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.016314867231994868, + "grad_norm": 0.11421401053667068, + "kept_token_count": 736.265625, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9912353754043579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008764622639864683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.30101013183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7981262207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5350, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.0999168187379837, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.01699622571468353, + "grad_norm": 0.11675813794136047, + "kept_token_count": 736.396875, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9908154010772705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009184591937810183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.83733825683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9361267089844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5360, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.10006804615259171, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.01617362592369318, + "grad_norm": 0.12153837084770203, + "kept_token_count": 736.275, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9907251298427582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009274855814874172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.59105224609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9884002685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5370, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.09972274303436279, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.017592441663146018, + "grad_norm": 0.11454000324010849, + "kept_token_count": 736.565625, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9912999391555786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00870004491880536, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.1998596191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.45786437988284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10010727420449257, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.018713366147130726, + "grad_norm": 0.11730758845806122, + "kept_token_count": 736.21875, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9907405853271485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009259420447051525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.8359436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6881988525391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.10014051720499992, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.01809489680454135, + "grad_norm": 0.12116117775440216, + "kept_token_count": 736.6703125, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9911677002906799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008832308370620013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.8494812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.68947143554686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5400, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.0993656761944294, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.017400308791548014, + "grad_norm": 0.12487519532442093, + "kept_token_count": 736.9484375, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9910008609294891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00899912156164646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.30402221679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4593872070312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09957463666796684, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.018001504056155682, + "grad_norm": 0.12081434577703476, + "kept_token_count": 736.6578125, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9912145495414734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008785457722842694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.13263244628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.253125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5420, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.09993706718087196, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.01672718543559313, + "grad_norm": 0.1304171234369278, + "kept_token_count": 736.446875, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9909303605556488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009069634694606065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.66407775878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0773956298828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5430, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.1003041259944439, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.015518013574182987, + "grad_norm": 0.1262182742357254, + "kept_token_count": 736.178125, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9918523907661438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008147604297846556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.8460235595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.05910034179686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5440, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10014982372522355, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.015904196258634328, + "grad_norm": 0.1591663956642151, + "kept_token_count": 736.2765625, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9909720718860626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009027919452637433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.0544769287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0172454833984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5450, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.10002916678786278, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.015569278970360756, + "grad_norm": 0.13192963600158691, + "kept_token_count": 736.5234375, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9912851095199585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008714896067976952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.4923492431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.37684936523436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5460, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09998720288276672, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.01751462873071432, + "grad_norm": 0.10379365086555481, + "kept_token_count": 736.3546875, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9908000409603119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009199956245720387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.10499572753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4722625732422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5470, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.10043553709983825, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.01721834121271968, + "grad_norm": 0.11544280499219894, + "kept_token_count": 736.1109375, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909179270267486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009082061145454646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.89133605957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.65443725585936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.09927734285593033, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.017812123987823725, + "grad_norm": 0.14625631272792816, + "kept_token_count": 737.153125, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9912532806396485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008746704924851656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.7934936523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8078247070313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5490, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.10018436387181281, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.0206086965277791, + "grad_norm": 0.09028328210115433, + "kept_token_count": 736.4609375, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.990855610370636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009144394006580115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.6518493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.2223175048828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.10041778907179832, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.015964803006500005, + "grad_norm": 0.12436536699533463, + "kept_token_count": 735.9734375, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9912190675735474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008780918829143048, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.6515411376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9809875488281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5510, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.10042824745178222, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.017453554458916186, + "grad_norm": 0.1034196987748146, + "kept_token_count": 735.9875, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9914917290210724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008508262317627668, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.3878173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9206787109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5520, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.09968102276325226, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.015154412481933832, + "grad_norm": 0.14823707938194275, + "kept_token_count": 736.63125, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9912079572677612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087920474819839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.6036071777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.50867919921876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5530, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.0996360644698143, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.01934999320656061, + "grad_norm": 0.14695963263511658, + "kept_token_count": 736.9515625, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9906469821929932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00935303308069706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.8251434326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9379913330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5540, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.09967220947146416, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.017427546344697474, + "grad_norm": 0.13114523887634277, + "kept_token_count": 736.5328125, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912808895111084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00871912445873022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.1363067626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8908233642578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5550, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.09985748678445816, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.020441491343080997, + "grad_norm": 0.11759302020072937, + "kept_token_count": 736.90625, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9911101520061493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008889839611947536, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.2139099121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.5517883300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5560, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.09998610019683837, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.017902766726911067, + "grad_norm": 0.13450050354003906, + "kept_token_count": 736.775, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913953185081482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008604671619832516, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8866027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.16631469726565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5570, + "valid_token_count": 818.6265625 + }, + { + "actual_token_mask_ratio": 0.09945809468626976, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.020854093413800002, + "grad_norm": 0.11947889626026154, + "kept_token_count": 737.2203125, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.99087553024292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009124475903809071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.0457458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.73097534179686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5580, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.10022733360528946, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.01745673716068268, + "grad_norm": 0.12120834738016129, + "kept_token_count": 736.165625, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9911664485931396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008833546191453934, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.36582336425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.27178649902345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.10048759952187539, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.02056675273925066, + "grad_norm": 0.1134575605392456, + "kept_token_count": 736.2734375, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9914570093154907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008543012384325265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.61939392089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7480163574219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5600, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.10013868138194085, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.019131498783826827, + "grad_norm": 0.09597063809633255, + "kept_token_count": 736.5046875, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9908598780632019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009140110854059458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.72333374023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3440368652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09998843297362328, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.01743866326287389, + "grad_norm": 0.10467395931482315, + "kept_token_count": 736.28125, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912041902542115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008795808721333741, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.09986267089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9175323486328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5620, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.09982202872633934, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.020880436431616543, + "grad_norm": 0.13511717319488525, + "kept_token_count": 736.628125, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9914591014385223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854091290384531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.3429260253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.09605712890624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5630, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.09922620505094529, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.016394533682614565, + "grad_norm": 0.1330314725637436, + "kept_token_count": 737.034375, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9910772562026977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00892274584621191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.6343078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.26214294433595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5640, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09988313317298889, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.019270382728427647, + "grad_norm": 0.13251861929893494, + "kept_token_count": 736.7921875, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9903654217720032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009634580835700035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.8332763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.7906524658203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5650, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.1003791257739067, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.01700556492432952, + "grad_norm": 0.1176944226026535, + "kept_token_count": 735.9515625, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9902852594852447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009714735858142375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.6931854248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.77831115722654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5660, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.10015799477696419, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.019994173571467398, + "grad_norm": 0.15074510872364044, + "kept_token_count": 736.1296875, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9908320367336273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009167957119643688, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.33690185546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6001678466797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5670, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.10013407468795776, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.015809189155697824, + "grad_norm": 0.13614770770072937, + "kept_token_count": 736.3234375, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9910130977630616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008986893761903047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.73822631835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.94561462402345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5680, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09999942034482956, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.01992628676816821, + "grad_norm": 0.13903361558914185, + "kept_token_count": 736.3078125, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9906838655471801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009316131379455328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.67737731933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8748260498047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10068855583667755, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.014989173877984286, + "grad_norm": 0.10931187868118286, + "kept_token_count": 735.96875, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9907092213630676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009290781524032354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.0346343994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.47193908691406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5700, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.10093496963381768, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.019656083639711143, + "grad_norm": 0.1070186048746109, + "kept_token_count": 735.95625, + "learning_rate": 2.841428725613801e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9912579834461213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008742010965943337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.90390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3732421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5710, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.10000007152557373, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.018600970786064864, + "grad_norm": 0.13313129544258118, + "kept_token_count": 736.4859375, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914106786251068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008589321281760931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.0677154541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8330383300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10004592835903167, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.022584995627403258, + "grad_norm": 0.13366104662418365, + "kept_token_count": 736.2171875, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0272, + "mgd_cos_sim": 0.9907541573047638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009245847910642624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.4568786621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4129241943359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5730, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.10067608952522278, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.018463741801679135, + "grad_norm": 0.13797876238822937, + "kept_token_count": 735.75, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9909072816371918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009092714358121156, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.8814208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.24700317382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.09975195750594139, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.016649857722222806, + "grad_norm": 0.1138969287276268, + "kept_token_count": 736.7921875, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912719845771789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008728026691824198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.08258056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3330047607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5750, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.10004743114113808, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.016110673639923336, + "grad_norm": 0.13543450832366943, + "kept_token_count": 736.1546875, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9913949251174927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008605071809142828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.20623168945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.82196044921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5760, + "valid_token_count": 817.9921875 + }, + { + "actual_token_mask_ratio": 0.09989047646522523, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.017485420033335686, + "grad_norm": 0.0912153497338295, + "kept_token_count": 736.390625, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9917951166629791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008204879891127347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.98089294433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.79819641113284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5770, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09898069500923157, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.019407017342746257, + "grad_norm": 0.1321977823972702, + "kept_token_count": 737.190625, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.991254460811615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00874553583562374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.8088104248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.1840362548828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5780, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.10027987957000732, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.015309407375752926, + "grad_norm": 0.09129631519317627, + "kept_token_count": 736.459375, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.99157634973526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008423651196062565, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.23682861328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8031066894531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5790, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.10009967908263206, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.016116688679903745, + "grad_norm": 0.11761844903230667, + "kept_token_count": 736.2171875, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9912148416042328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008785162027925253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.6845306396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.00925598144534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5800, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.09926969185471535, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.016532658506184815, + "grad_norm": 0.11466272175312042, + "kept_token_count": 736.909375, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9911467432975769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008853244874626399, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.99571228027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.33515014648435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5810, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10033827275037766, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.01573074981570244, + "grad_norm": 0.09376701712608337, + "kept_token_count": 736.0703125, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9913291156291961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008670896012336015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.40668334960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6558807373047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5820, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.10050920695066452, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.017124150041490795, + "grad_norm": 0.10068732500076294, + "kept_token_count": 735.8609375, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9914350807666779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008564926218241453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.2965545654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.770068359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5830, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.10021234601736069, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.01616004789248109, + "grad_norm": 0.16438277065753937, + "kept_token_count": 736.28125, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9914324462413788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008567544631659985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.74811401367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.04542846679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5840, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.10001455023884773, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.01705875722691417, + "grad_norm": 0.13575983047485352, + "kept_token_count": 736.2875, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9913312256336212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008668768126517535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.44561157226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.99537963867186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5850, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.09972611889243126, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.016463339515030383, + "grad_norm": 0.10163264721632004, + "kept_token_count": 736.378125, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9913035988807678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008696394320577384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.75783386230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5457458496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5860, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.100413628667593, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.016366658080369235, + "grad_norm": 0.12651970982551575, + "kept_token_count": 735.9890625, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9909022748470306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009097733348608018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.99803466796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1056762695313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5870, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.09989387318491935, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.017766637448221444, + "grad_norm": 0.1171388030052185, + "kept_token_count": 736.3921875, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9909906327724457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009009362664073706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.2887725830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9804748535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5880, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.1003777340054512, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.014746779017150402, + "grad_norm": 0.11287898570299149, + "kept_token_count": 736.3421875, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9913525521755219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008647448662668467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.5570526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7139923095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5890, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.09987010955810546, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.016458909120410682, + "grad_norm": 0.08371839672327042, + "kept_token_count": 736.575, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9914967358112335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008503256645053626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.6417602539062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4404266357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5900, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.09981149584054946, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.016632204595953225, + "grad_norm": 0.12088877707719803, + "kept_token_count": 736.35, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9915499866008759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008450021594762802, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.0506958007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6770446777344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.09952883049845695, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.01702773040160537, + "grad_norm": 0.12915737926959991, + "kept_token_count": 736.915625, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909423053264618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009057691879570484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.934228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1039245605469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5920, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.09920377731323242, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.019729688111692668, + "grad_norm": 0.155594602227211, + "kept_token_count": 737.23125, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.990654981136322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009345009643584489, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.1744750976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2381896972656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5930, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.10012697651982308, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.017983109503984452, + "grad_norm": 0.1247003823518753, + "kept_token_count": 736.31875, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9908759951591491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00912401219829917, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.5653533935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1942413330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5940, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.09997465312480927, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.019438605383038522, + "grad_norm": 0.13301332294940948, + "kept_token_count": 736.565625, + "learning_rate": 2.823203278868685e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9909390211105347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009060977771878243, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.2045928955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.13670654296874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5950, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.10064128339290619, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.016964344680309294, + "grad_norm": 0.1534300148487091, + "kept_token_count": 736.18125, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9914975583553314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850243316963315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.7957733154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5570343017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5960, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.10008811801671982, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.016837163548916578, + "grad_norm": 0.1370839625597, + "kept_token_count": 736.3515625, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918895363807678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008110443502664566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.6208435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.78713073730466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5970, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.09977182149887084, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.01580086974427104, + "grad_norm": 0.12642310559749603, + "kept_token_count": 736.8703125, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9917333781719208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008266620058566331, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.36558837890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1505462646484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5980, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.10049076676368714, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.01751555232331157, + "grad_norm": 0.12925148010253906, + "kept_token_count": 736.215625, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9920124530792236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007987551391124725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.3216979980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0658325195312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 5990, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09933066815137863, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.01612066337838769, + "grad_norm": 0.13822118937969208, + "kept_token_count": 737.3015625, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9910656154155731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008934373129159212, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.58152160644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.75928955078126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6000, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.09975197315216064, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.018330549634993077, + "grad_norm": 0.14866706728935242, + "kept_token_count": 736.38125, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9913747906684875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008625196013599633, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.6378204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7635467529297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6010, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.0998687706887722, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.017353529296815395, + "grad_norm": 0.12557871639728546, + "kept_token_count": 736.5546875, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913461983203888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008653800189495086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.3710632324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.82227783203126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6020, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.09958380609750747, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.019627130590379238, + "grad_norm": 0.13447034358978271, + "kept_token_count": 736.6625, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9912313520908356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008768647257238627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9375762939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.66026306152344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6030, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.09975801855325699, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.015552387945353985, + "grad_norm": 0.14512325823307037, + "kept_token_count": 736.284375, + "learning_rate": 2.816129283598112e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9911226511001587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008877343777567149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.38164978027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0638061523438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6040, + "valid_token_count": 817.86875 + }, + { + "actual_token_mask_ratio": 0.10071708112955094, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01797550329938531, + "grad_norm": 0.10428348928689957, + "kept_token_count": 735.975, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9912312626838684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008768740389496089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.0251770019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.6994659423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6050, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09921058341860771, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.01694457922130823, + "grad_norm": 0.13052502274513245, + "kept_token_count": 737.2171875, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9909317374229432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009068267606198787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.6597137451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.55689392089846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6060, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.10036164745688439, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.016609217412769793, + "grad_norm": 0.09580937027931213, + "kept_token_count": 736.409375, + "learning_rate": 2.813742464676556e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912357330322266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008764252811670304, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.36948547363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.78416137695314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6070, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.0993713915348053, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.02110506845638156, + "grad_norm": 0.11088831722736359, + "kept_token_count": 736.6703125, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9914504289627075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854957727715373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.0434936523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.78261108398436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6080, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.09994083270430565, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.017983285430818795, + "grad_norm": 0.15295782685279846, + "kept_token_count": 736.3296875, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9912751972675323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008724791649729013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.8110595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.57591247558594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6090, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09919478371739388, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.017470745369791985, + "grad_norm": 0.1240386962890625, + "kept_token_count": 737.28125, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9915968656539917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008403140120208263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.55116882324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.48047790527346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6100, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.09961042702198028, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.01971552213653922, + "grad_norm": 0.1346360445022583, + "kept_token_count": 737.3515625, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9907821595668793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009217835217714309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.2855590820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3975311279297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6110, + "valid_token_count": 818.9234375 + }, + { + "actual_token_mask_ratio": 0.09994789212942123, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.01863914802670479, + "grad_norm": 0.1334911286830902, + "kept_token_count": 736.5484375, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9911498248577117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008850163966417312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.9871124267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9084228515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6120, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.09944064319133758, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.018554149195551872, + "grad_norm": 0.11896727234125137, + "kept_token_count": 736.66875, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9913102328777313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008689765818417072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.54848022460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4809265136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6130, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.09943063408136368, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.0186589234508574, + "grad_norm": 0.15759535133838654, + "kept_token_count": 736.7359375, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914705872535705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008529420290142298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.8561492919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.2919586181641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6140, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.10007625445723534, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.015385706163942813, + "grad_norm": 0.10600320994853973, + "kept_token_count": 736.4125, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9915959656238555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008404044900089502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.45624084472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.57225646972654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6150, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.09976979717612267, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.017596199084073304, + "grad_norm": 0.16070808470249176, + "kept_token_count": 736.4625, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908503890037537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009149609599262476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.5562438964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.4318572998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.10034526437520981, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.015499076060950755, + "grad_norm": 0.1204390823841095, + "kept_token_count": 736.146875, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9915898382663727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008410166949033737, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.79857177734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2077209472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6170, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.09958984926342965, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.016354963555932044, + "grad_norm": 0.1318112015724182, + "kept_token_count": 736.9078125, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9906949698925018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009305022470653058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.06593322753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1023895263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6180, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.10060372054576874, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.017095868568867446, + "grad_norm": 0.12789320945739746, + "kept_token_count": 735.8375, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9911868333816528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008813159260898828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.7636016845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.1193328857422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6190, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.10055295825004577, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.01741789057850838, + "grad_norm": 0.15281538665294647, + "kept_token_count": 736.06875, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9914628207683563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0085371776483953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.6201934814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4745147705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6200, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.10034728050231934, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.017495868913829326, + "grad_norm": 0.1025235652923584, + "kept_token_count": 736.03125, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.991365772485733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00863422192633152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.313916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.6600738525391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6210, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.099708703905344, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.016151005681604148, + "grad_norm": 0.0843350887298584, + "kept_token_count": 736.65625, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9914229273796081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008577065076678991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.2892639160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.02642517089845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6220, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.10054821372032166, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.016414770763367413, + "grad_norm": 0.14657476544380188, + "kept_token_count": 736.1203125, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9910213708877563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008978634141385555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.06263427734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.758837890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.09981984123587609, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.0178810628131032, + "grad_norm": 0.13937771320343018, + "kept_token_count": 736.321875, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9917557716369629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008244230970740318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.26646118164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.33897705078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6240, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.09973321184515953, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.01663856003433466, + "grad_norm": 0.1399214267730713, + "kept_token_count": 736.7828125, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9917402148246766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00825977921485901, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.6505828857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6023468017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6250, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.10009165704250336, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.016475459467619656, + "grad_norm": 0.13445839285850525, + "kept_token_count": 736.4609375, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9915636301040649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00843637716025114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.84966735839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.65472717285155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6260, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09946850165724755, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.01592831164598465, + "grad_norm": 0.128105029463768, + "kept_token_count": 736.753125, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9917450070381164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008254987001419068, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.67872009277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.57696228027345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6270, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.10062332227826118, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.015094349160790443, + "grad_norm": 0.11083389818668365, + "kept_token_count": 735.70625, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9911786079406738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008821397274732589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.2504455566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.38677978515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6280, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.10046997293829918, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.018347966391593216, + "grad_norm": 0.12355305254459381, + "kept_token_count": 736.353125, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.991014051437378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008985949959605933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.67078552246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.9408325195312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6290, + "valid_token_count": 818.596875 + }, + { + "actual_token_mask_ratio": 0.09978966787457466, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.016754762921482324, + "grad_norm": 0.09062041342258453, + "kept_token_count": 736.553125, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9917087078094482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00829130234196782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.9083953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.56995544433596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6300, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.10009758025407792, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.015784479957073928, + "grad_norm": 0.10344633460044861, + "kept_token_count": 736.45625, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9918272256851196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008172763977199793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.82684631347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.9321319580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6310, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09982735142111779, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.0173728764988482, + "grad_norm": 0.11968841403722763, + "kept_token_count": 736.5984375, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9910647869110107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00893521700054407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.6676025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.6123352050781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6320, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.09968462809920312, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.018764102645218373, + "grad_norm": 0.1394425928592682, + "kept_token_count": 736.4953125, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9910739839076996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008926016185432673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6061553955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3318695068359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6330, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.09987755641341209, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.016469830088317393, + "grad_norm": 0.12304830551147461, + "kept_token_count": 736.7828125, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9913209319114685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0086790693923831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.63462524414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.79111328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6340, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.09987786710262299, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.020232310704886913, + "grad_norm": 0.12059356272220612, + "kept_token_count": 736.4890625, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9915919125080108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008408088609576226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.9733459472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.94140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.0994420014321804, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.01604714011773467, + "grad_norm": 0.12900757789611816, + "kept_token_count": 736.9109375, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.99119633436203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008803666941821576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.5057373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.32740173339846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6360, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.1003642700612545, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.016892096493393183, + "grad_norm": 0.1368710696697235, + "kept_token_count": 736.0890625, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917616307735443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008238357119262218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.79820861816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.0179412841797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6370, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.10014643371105195, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.01690262332558632, + "grad_norm": 0.11996092647314072, + "kept_token_count": 736.7578125, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9907343983650208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009265604056417942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.6825256347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1421142578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6380, + "valid_token_count": 818.7484375 + }, + { + "actual_token_mask_ratio": 0.09984666630625724, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.017912443727254868, + "grad_norm": 0.12296651303768158, + "kept_token_count": 736.5375, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9912940323352813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00870596319437027, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.698388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.89482421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6390, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.1005137287080288, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.01993437148630619, + "grad_norm": 0.14485634863376617, + "kept_token_count": 735.95, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9904806256294251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009519372135400772, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.7719451904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.1260223388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6400, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.09956740513443947, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.019299892242997885, + "grad_norm": 0.19276413321495056, + "kept_token_count": 736.7890625, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9914656519889832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008534345589578152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.9848358154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.7104461669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09989107102155685, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.01810712171718478, + "grad_norm": 0.10484922677278519, + "kept_token_count": 736.3765625, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9913764357566833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008623571041971445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.6919006347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.21925964355466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6420, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.09945369735360146, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.016549620497971772, + "grad_norm": 0.11958251148462296, + "kept_token_count": 736.8171875, + "learning_rate": 2.783987643028595e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9910693287849426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008930671587586403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.0793029785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.4716766357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6430, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.09956732988357545, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.018670466914772987, + "grad_norm": 0.12394986301660538, + "kept_token_count": 736.7203125, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9909202337265015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009079763107001782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.1002624511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.5678771972656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6440, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.09997517615556717, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.01787214521318674, + "grad_norm": 0.11418414115905762, + "kept_token_count": 736.3671875, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9912542998790741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008745692484080792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.4113830566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.0242950439453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09984472766518593, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.018115720991045237, + "grad_norm": 0.1926196962594986, + "kept_token_count": 736.521875, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9914397895336151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008560217451304197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.23033447265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.774755859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6460, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.10026193857192993, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.015855829883366823, + "grad_norm": 0.13325373828411102, + "kept_token_count": 736.371875, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9915739476680756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008426049444824456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.68568115234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.63364562988284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6470, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.10082449316978455, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.01654763938859105, + "grad_norm": 0.117572121322155, + "kept_token_count": 735.9453125, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919391572475433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008060839399695397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.3810668945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.32470703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6480, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.10020555928349495, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.019370168354362248, + "grad_norm": 0.11201975494623184, + "kept_token_count": 736.30625, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.991412878036499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008587120566517115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.3271209716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.3763427734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6490, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.09962309300899505, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.01819578567519784, + "grad_norm": 0.11842924356460571, + "kept_token_count": 736.5546875, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9910086810588836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008991308137774467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.83372497558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.9825103759766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6500, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.09930138140916825, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.01890970319509506, + "grad_norm": 0.09727782756090164, + "kept_token_count": 737.009375, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9911151945590972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008884802926331758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.0572204589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.14747314453126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6510, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10008958578109742, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.017099815234541894, + "grad_norm": 0.09651400148868561, + "kept_token_count": 736.39375, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909726560115815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009027336724102497, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.78298034667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.85517578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6520, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.10053271874785423, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.01746214060112834, + "grad_norm": 0.12562832236289978, + "kept_token_count": 735.6171875, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912752032279968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00872479910030961, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.5995330810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.3990112304688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6530, + "valid_token_count": 817.8359375 + }, + { + "actual_token_mask_ratio": 0.10031402707099915, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.015225390624254941, + "grad_norm": 0.125686913728714, + "kept_token_count": 736.1765625, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9913537442684174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008646263927221298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.15142822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.3091705322266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6540, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09990351125597954, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.01680394131690264, + "grad_norm": 0.11689392477273941, + "kept_token_count": 736.165625, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9916404783725739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008359523303806782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.9462158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.28985290527345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6550, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.0997971199452877, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.017122523766011, + "grad_norm": 0.1280042976140976, + "kept_token_count": 736.725, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9912181377410889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008781862445175648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.0674682617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.35675354003905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6560, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09994725659489631, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.017817832063883543, + "grad_norm": 0.09729477763175964, + "kept_token_count": 736.54375, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913165390491485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008683468215167522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.1614288330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.152783203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6570, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10002192333340645, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.015442897379398347, + "grad_norm": 0.1067785695195198, + "kept_token_count": 736.425, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9905796825885773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009420313313603402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.8245513916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.80275573730466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6580, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09969042837619782, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.01773520829156041, + "grad_norm": 0.10777966678142548, + "kept_token_count": 736.3875, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9914704620838165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008529539871960878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.9678192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1498199462891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6590, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.10045696273446084, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.018980497494339943, + "grad_norm": 0.13297846913337708, + "kept_token_count": 736.16875, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9907126188278198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009287386294454336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.610400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.2025970458984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6600, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.10065192952752114, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.01806079288944602, + "grad_norm": 0.10672620683908463, + "kept_token_count": 735.790625, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915515840053558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008448415715247393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.1177093505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4640869140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.09988859668374062, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.019530014321208, + "grad_norm": 0.15642701089382172, + "kept_token_count": 736.484375, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9909231185913085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00907688345760107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.52953186035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1579528808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6620, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.09937120154500008, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.01951832212507725, + "grad_norm": 0.1281348019838333, + "kept_token_count": 736.909375, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9916909813880921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008309013210237026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.8479858398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1628662109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6630, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.10044568553566932, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.01951263388618827, + "grad_norm": 0.13767021894454956, + "kept_token_count": 736.409375, + "learning_rate": 2.765693889036747e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9910560548305511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008943939581513405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.20765686035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.0243225097656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6640, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.09984293133020401, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.0215216089040041, + "grad_norm": 0.11662312597036362, + "kept_token_count": 736.5828125, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9909116685390472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009088328294456004, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.30238647460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.5631591796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6650, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.1005672425031662, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.015862797759473322, + "grad_norm": 0.12622246146202087, + "kept_token_count": 735.9171875, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.991020143032074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00897986302152276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.99046020507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.1580413818359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6660, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09939537942409515, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.018829365633428095, + "grad_norm": 0.11919349431991577, + "kept_token_count": 736.903125, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9913950562477112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008604937605559826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.4587768554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.24677124023435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6670, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09968896135687828, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.01491152960807085, + "grad_norm": 0.10213082283735275, + "kept_token_count": 736.590625, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9915974318981171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0084025620482862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.34795532226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.8930633544922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6680, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10007000416517257, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.01593266874551773, + "grad_norm": 0.1248193308711052, + "kept_token_count": 736.2375, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9918190002441406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00818100105971098, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.4901062011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3440795898438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6690, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.0997833639383316, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.0172868344001472, + "grad_norm": 0.12910479307174683, + "kept_token_count": 736.7421875, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9912420153617859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008757987339049578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.1254638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.68402099609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6700, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.099662084877491, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.018777713272720577, + "grad_norm": 0.10663257539272308, + "kept_token_count": 736.89375, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.990599125623703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009400875121355057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.59853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.7195587158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6710, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.100367970764637, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.020477460231631994, + "grad_norm": 0.12001487612724304, + "kept_token_count": 736.4703125, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9914293169975281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008570663910359145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.22272338867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.86029357910155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.10019954144954682, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.018189784418791533, + "grad_norm": 0.1027078926563263, + "kept_token_count": 736.303125, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9916861057281494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008313899394124746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.80426635742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.84421997070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6730, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.0995868667960167, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.017175370641052724, + "grad_norm": 0.11706923693418503, + "kept_token_count": 736.5046875, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9912407219409942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008759278617799283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9688415527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.99449462890624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6740, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.09986350983381272, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.019900398422032596, + "grad_norm": 0.1224297285079956, + "kept_token_count": 736.740625, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9911652684211731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008834733255207538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.02898559570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.94537353515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6750, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.09996725171804428, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.0189097311347723, + "grad_norm": 0.1462908238172531, + "kept_token_count": 736.4359375, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916406214237213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008359363954514264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.6768737792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.50169677734374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6760, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.10021542534232139, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.018107380624860524, + "grad_norm": 0.12946677207946777, + "kept_token_count": 736.259375, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9912143051624298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008785695303231478, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.05450744628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.9109802246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6770, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10007994696497917, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.018596986774355173, + "grad_norm": 0.1293887346982956, + "kept_token_count": 736.16875, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914394915103912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008560520503669978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.8800567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.09250793457034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6780, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.09988356530666351, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.017756818048655986, + "grad_norm": 0.13445894420146942, + "kept_token_count": 736.553125, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9913323998451233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00866759680211544, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.8444427490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.2303405761719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6790, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.09973222464323044, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.01732054715976119, + "grad_norm": 0.14740052819252014, + "kept_token_count": 736.565625, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9913455545902252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008654436003416777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.03034057617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.8824005126953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6800, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09991474002599716, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.018932150956243277, + "grad_norm": 0.11851662397384644, + "kept_token_count": 736.8140625, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9910260796546936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008973903860896826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.40135803222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.1987548828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6810, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.10026091188192368, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.01983129484578967, + "grad_norm": 0.11422120034694672, + "kept_token_count": 736.078125, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9904022932052612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009597699716687202, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.26100158691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.4783996582031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6820, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10040879473090172, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.016194954235106706, + "grad_norm": 0.10242683440446854, + "kept_token_count": 735.8828125, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917339324951172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008266065176576376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.7554107666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.56788024902346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6830, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.10050481483340264, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.018144808989018203, + "grad_norm": 0.11925660818815231, + "kept_token_count": 736.13125, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.990742701292038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009257288370281457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.2084014892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.2648071289062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6840, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.09999859184026719, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.01683491189032793, + "grad_norm": 0.11112993955612183, + "kept_token_count": 736.328125, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9911657571792603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008834247384220362, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.809326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.6451812744141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6850, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.09935620129108429, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.016665334161370993, + "grad_norm": 0.10844007879495621, + "kept_token_count": 737.0328125, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912467122077941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00875329403206706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.3032562255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.4920196533203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6860, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.10030995458364486, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.01697041094303131, + "grad_norm": 0.1487833708524704, + "kept_token_count": 736.1609375, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9918114423751831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008188569080084562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.0474365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.2549621582031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6870, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09979571178555488, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.018070801999419926, + "grad_norm": 0.14391224086284637, + "kept_token_count": 736.6140625, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9916837155818939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008316289726644754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.6521514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.4204620361328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6880, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.10010728687047958, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.017003103625029325, + "grad_norm": 0.10705523192882538, + "kept_token_count": 736.18125, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.991532826423645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008467167150229216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.1869812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.4760406494141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6890, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.10004512295126915, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.01753057390451431, + "grad_norm": 0.11415207386016846, + "kept_token_count": 736.5328125, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9917464911937713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008253521751612425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.59987182617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.937744140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6900, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10005684345960617, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.018586464039981366, + "grad_norm": 0.12463260442018509, + "kept_token_count": 736.2640625, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9905811905860901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009418813232332467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.4275207519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.2319061279297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6910, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.10068273097276688, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01727646393701434, + "grad_norm": 0.10490614175796509, + "kept_token_count": 735.7453125, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9917385876178741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261407259851694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.1008239746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.37926330566404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6920, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.09985717609524727, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.018505755998194217, + "grad_norm": 0.14081358909606934, + "kept_token_count": 736.7046875, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9914576053619385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854238523170352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.5310577392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.6831756591797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6930, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.09945166185498237, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.016473003663122654, + "grad_norm": 0.11191376298666, + "kept_token_count": 737.0671875, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9908462941646576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00915370499715209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.60997314453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.2151824951172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6940, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.09960720017552376, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.018462602328509093, + "grad_norm": 0.13457749783992767, + "kept_token_count": 736.9, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9918354570865631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00816455939784646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.92823486328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.5319000244141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6950, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.09964993670582771, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.018212056159973143, + "grad_norm": 0.13605079054832458, + "kept_token_count": 736.7, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9914240598678589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008575930818915367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.75665893554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.11334533691405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6960, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.09959184601902962, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.017811419907957315, + "grad_norm": 0.09846873581409454, + "kept_token_count": 736.8359375, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9911401391029357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008859851770102978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.85972900390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.30491943359374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1000936321914196, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.01863744305446744, + "grad_norm": 0.10064076632261276, + "kept_token_count": 736.6046875, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9918422341346741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157779928296804, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.1608489990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8608337402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6980, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.09980174303054809, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.017707858048379422, + "grad_norm": 0.11936593055725098, + "kept_token_count": 736.5515625, + "learning_rate": 2.733702192878577e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9913587927818298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008641207311302423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.3555969238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.7054748535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 6990, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.0999696634709835, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.01674727229401469, + "grad_norm": 0.09689986705780029, + "kept_token_count": 736.665625, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9908157646656036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00918424418196082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.33599853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.3912628173828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7000, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.10003130808472634, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01877432158216834, + "grad_norm": 0.13482694327831268, + "kept_token_count": 736.275, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9910930514335632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008906958904117347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.5925018310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.55535583496095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7010, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.10020247101783752, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.016140519734472038, + "grad_norm": 0.08168833702802658, + "kept_token_count": 736.359375, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9914817154407501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008518276922404765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.7013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.7041015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7020, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.09917979091405868, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.0158777367323637, + "grad_norm": 0.1272611767053604, + "kept_token_count": 737.1671875, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9910398721694946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008960142638534308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.7233947753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.71221618652345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7030, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.09987900778651237, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.017701573390513658, + "grad_norm": 0.09910275787115097, + "kept_token_count": 736.453125, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9916351675987244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008364816848188638, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.6065277099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.32281494140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7040, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.10005944892764092, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.01759485639631748, + "grad_norm": 0.12098272144794464, + "kept_token_count": 736.146875, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9914207935333252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008579211123287678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.943408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.6318084716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7050, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.10010690316557884, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.015978926699608563, + "grad_norm": 0.14519034326076508, + "kept_token_count": 736.2375, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9917321979999543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008267788775265216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.3574615478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.547802734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7060, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.09974130392074584, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.019164147879928352, + "grad_norm": 0.12898871302604675, + "kept_token_count": 736.778125, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9907545506954193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009245438128709793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.7418182373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.4487640380859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7070, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10068175941705704, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.01594393439590931, + "grad_norm": 0.10585497319698334, + "kept_token_count": 735.6359375, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9916184902191162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00838149581104517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.22251892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.8784210205078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.1001988835632801, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.017476065177470447, + "grad_norm": 0.15314915776252747, + "kept_token_count": 736.4421875, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9917689740657807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008231011126190424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.7767730712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.5646026611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7090, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.10026469677686692, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.016527409385889768, + "grad_norm": 0.11079203337430954, + "kept_token_count": 736.128125, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9912665247917175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008733472600579262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.63941345214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.4292938232422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7100, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.09997257217764854, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.018205766379833222, + "grad_norm": 0.09358378499746323, + "kept_token_count": 736.2890625, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9917504966259003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008249510265886783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.73968200683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.0585540771484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7110, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.09974186718463898, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.017593548446893693, + "grad_norm": 0.1225639283657074, + "kept_token_count": 736.534375, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.991514015197754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008485974278301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.5094299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.24501037597656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7120, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.09999545440077781, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.018086678721010684, + "grad_norm": 0.10682636499404907, + "kept_token_count": 736.3046875, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915842056274414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008415797352790832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.9942260742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.2537475585938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7130, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.09978434592485427, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.016700176522135735, + "grad_norm": 0.13254164159297943, + "kept_token_count": 736.5203125, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9914535164833069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008546489384025335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.75476989746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.3363372802734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.09997461885213851, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.014781683124601841, + "grad_norm": 0.11367420107126236, + "kept_token_count": 736.3953125, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9913172662258148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008682724274694919, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.37351379394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.24969482421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7150, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.10025931149721146, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.017532260343432426, + "grad_norm": 0.13863158226013184, + "kept_token_count": 736.2953125, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9916303038597107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00836969455704093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.35789794921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.6093353271484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7160, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.10043763220310212, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.018683572206646203, + "grad_norm": 0.12133979052305222, + "kept_token_count": 736.3265625, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918306291103363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008169366605579853, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.47474975585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5424102783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7170, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.0999815121293068, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.01744186906144023, + "grad_norm": 0.12601344287395477, + "kept_token_count": 736.515625, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9915726363658905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008427365124225617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.55783996582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3426910400391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7180, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.0997083380818367, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.01790472064167261, + "grad_norm": 0.10587607324123383, + "kept_token_count": 737.0328125, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9911858677864075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008814132958650588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.4861145019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.956103515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7190, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.09949239566922188, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.01623760340735316, + "grad_norm": 0.09734756499528885, + "kept_token_count": 736.9109375, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9913313925266266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008668611012399196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.31092224121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.7017822265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7200, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.10031873509287834, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.01709457067772746, + "grad_norm": 0.12307875603437424, + "kept_token_count": 736.0828125, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9911610722541809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00883892634883523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.67348022460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.74632873535154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7210, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.10065265968441964, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015646683424711226, + "grad_norm": 0.11412718147039413, + "kept_token_count": 736.1234375, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9909145653247833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009085431415587663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.96600952148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.57716979980466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7220, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.10002941861748696, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.018342688027769328, + "grad_norm": 0.11519049108028412, + "kept_token_count": 736.275, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9919918119907379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008008191362023354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.93517150878904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.0504211425781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7230, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.099334017932415, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.019372746162116528, + "grad_norm": 0.11712028831243515, + "kept_token_count": 737.0671875, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9916027188301086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008397294860333204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.10143127441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.17316284179685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7240, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10033400654792786, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.015067236311733723, + "grad_norm": 0.087351493537426, + "kept_token_count": 736.315625, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9921825051307678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007817485835403204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.382763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.81964111328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7250, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.09973201528191566, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.015593934711068869, + "grad_norm": 0.11658656597137451, + "kept_token_count": 736.65, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9917871832847596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008212804328650236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.31251220703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.08094177246096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7260, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.1005699135363102, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.018050095345824958, + "grad_norm": 0.11728038638830185, + "kept_token_count": 735.815625, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9913562178611756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008643776644021273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.5509826660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.2159851074219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7270, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.10028347298502922, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.017779705952852963, + "grad_norm": 0.13545416295528412, + "kept_token_count": 736.3, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.991266518831253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008733490761369467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.7930969238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.12599182128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7280, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09968329221010208, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.0187493565492332, + "grad_norm": 0.12457995116710663, + "kept_token_count": 736.5921875, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.990938687324524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009061316587030887, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.0347961425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.1552215576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7290, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.09948408678174019, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.01691943909972906, + "grad_norm": 0.1160818487405777, + "kept_token_count": 736.815625, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9918428599834442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157137036323547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.8924102783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.7994079589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7300, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09986195787787437, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.01840642634779215, + "grad_norm": 0.13442480564117432, + "kept_token_count": 736.740625, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9914244055747986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008575598057359458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.4817840576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.6338623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7310, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.09987839162349701, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.019729171972721816, + "grad_norm": 0.10834481567144394, + "kept_token_count": 736.4296875, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.991613358259201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386650681495666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.16502685546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.85186157226565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7320, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.09934904053807259, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.016525679919868707, + "grad_norm": 0.10278733819723129, + "kept_token_count": 736.628125, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9913416624069213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008658319246023893, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.5918334960937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.2725555419922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7330, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.09971084222197532, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.017551398649811745, + "grad_norm": 0.12347067892551422, + "kept_token_count": 736.859375, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9905895888805389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0094104016199708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.1098388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.26568298339845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7340, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.09928204268217086, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.016291130613535643, + "grad_norm": 0.11372546851634979, + "kept_token_count": 737.1734375, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9909927308559418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009007268212735653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.2808837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.214111328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.1003036729991436, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.021509499847888948, + "grad_norm": 0.10953821986913681, + "kept_token_count": 736.1515625, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9906962811946869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009303714800626039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.04893493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.3048797607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7360, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09969725534319877, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.02085420163348317, + "grad_norm": 0.12730015814304352, + "kept_token_count": 736.765625, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9906789839267731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009321012906730175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.4192779541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.526708984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7370, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09968970194458962, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.017536604404449464, + "grad_norm": 0.13546092808246613, + "kept_token_count": 736.834375, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910840392112732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008915958553552627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.75146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.22890014648436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7380, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.10099481120705604, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.01939082071185112, + "grad_norm": 0.13653714954853058, + "kept_token_count": 735.821875, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9914771795272828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008522828202694654, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.85479431152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.1186492919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.10036736205220223, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.016287724673748016, + "grad_norm": 0.11748749017715454, + "kept_token_count": 736.2109375, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9911880612373352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008811947889626026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6934020996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.86742248535154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7400, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.10029933676123619, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.016336483974009754, + "grad_norm": 0.11400984227657318, + "kept_token_count": 736.0640625, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9922676086425781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007732410822063685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.40404357910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.6843658447266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7410, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10038455948233604, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.01659477036446333, + "grad_norm": 0.09836950153112411, + "kept_token_count": 736.1875, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919367253780365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008063281886279583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0645263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.44212951660154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7420, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.10027061998844147, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.01694838497787714, + "grad_norm": 0.0968913584947586, + "kept_token_count": 736.4140625, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9913351476192475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008664849493652583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.00343627929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.1360626220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7430, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.10052193850278854, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.017077859211713075, + "grad_norm": 0.13919472694396973, + "kept_token_count": 735.95625, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9916133046150207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386698644608259, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.6345520019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.76202697753905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7440, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.10000099763274192, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.017148299422115088, + "grad_norm": 0.10706336796283722, + "kept_token_count": 736.51875, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.990879237651825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009120762255042792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.93019714355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.9734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7450, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.10056879520416259, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.018674427922815085, + "grad_norm": 0.13508734107017517, + "kept_token_count": 735.66875, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9917811155319214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008218884374946356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.0241943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.73211669921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7460, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.09981241077184677, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.017537047900259495, + "grad_norm": 0.10324271023273468, + "kept_token_count": 736.4015625, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9915822446346283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008417772967368365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.4652862548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.5829650878906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7470, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.10036915838718415, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.016076561249792575, + "grad_norm": 0.12838153541088104, + "kept_token_count": 736.403125, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9914896488189697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00851035825908184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.9163513183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.6547119140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7480, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.10035470128059387, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015069725550711155, + "grad_norm": 0.09017200022935867, + "kept_token_count": 736.134375, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9915273010730743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00847268821671605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.3801971435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.23919677734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7490, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10021963939070702, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.019019445683807135, + "grad_norm": 0.11872144788503647, + "kept_token_count": 736.2671875, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9909340977668762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009065913408994675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.24989013671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.61961059570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7500, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.09952419623732567, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.016116585303097963, + "grad_norm": 0.1267181932926178, + "kept_token_count": 736.8171875, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9907984495162964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009201550390571356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.60024108886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.5807586669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7510, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09975457340478897, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.01925890250131488, + "grad_norm": 0.12395837903022766, + "kept_token_count": 736.6, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9907342433929444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009265758283436298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.6368133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.97995910644534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7520, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.10059019699692726, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.016372206062078475, + "grad_norm": 0.12194227427244186, + "kept_token_count": 735.734375, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9913621604442596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008637848496437072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.3312957763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.24994506835935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.09938842579722404, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.015704306680709122, + "grad_norm": 0.11336351931095123, + "kept_token_count": 737.21875, + "learning_rate": 2.679739067375394e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9914379179477691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008562082331627608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.4679901123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.5801177978516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7540, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.09992112591862679, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.01589415781199932, + "grad_norm": 0.13842132687568665, + "kept_token_count": 736.915625, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9914182722568512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008581728395074605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.1168609619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.0611846923828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7550, + "valid_token_count": 818.71875 + }, + { + "actual_token_mask_ratio": 0.1007955938577652, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.01973288571462035, + "grad_norm": 0.15128742158412933, + "kept_token_count": 735.74375, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9912514567375184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008748536650091409, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.4220001220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.05103759765626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7560, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.10030655860900879, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.01576413493603468, + "grad_norm": 0.14549550414085388, + "kept_token_count": 736.3515625, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9916077077388763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008392290119081736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.1011505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.26897888183595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7570, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.09965973049402237, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.015295091643929481, + "grad_norm": 0.12475711852312088, + "kept_token_count": 736.2609375, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9914478182792663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008552172780036926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.5416961669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.0832122802734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7580, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.10023632869124413, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.01631406880915165, + "grad_norm": 0.11078283190727234, + "kept_token_count": 736.371875, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9917056858539581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294315543025732, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.1298553466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.61449279785154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7590, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.10054111927747726, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.01767542278394103, + "grad_norm": 0.13491931557655334, + "kept_token_count": 735.990625, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9916082978248596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008391715865582228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.20339050292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.1467651367187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7600, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10011475682258605, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.015805155597627164, + "grad_norm": 0.11630348116159439, + "kept_token_count": 736.434375, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.991447114944458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00855289287865162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.9200134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.05206909179685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7610, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.09977562204003335, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.01608005501329899, + "grad_norm": 0.09385215491056442, + "kept_token_count": 736.7078125, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922067642211914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007793243508785963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.3810211181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.6433837890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7620, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.1000951111316681, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.017362493090331554, + "grad_norm": 0.11296501755714417, + "kept_token_count": 736.44375, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915094196796417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008490591775625944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.4095825195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.8498046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7630, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.09988149479031563, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.017777725495398043, + "grad_norm": 0.116949662566185, + "kept_token_count": 736.928125, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9916538417339325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008346155192703009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.9016357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.2009552001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7640, + "valid_token_count": 818.7 + }, + { + "actual_token_mask_ratio": 0.0999048337340355, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.017267550993710756, + "grad_norm": 0.11328597366809845, + "kept_token_count": 736.35, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909922420978546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900775082409382, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.6630828857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.24456787109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7650, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.09984404146671295, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.020587138086557388, + "grad_norm": 0.16348253190517426, + "kept_token_count": 736.684375, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9911231458187103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008876845054328442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.31170959472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.1390686035156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7660, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09937980473041534, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.017374747525900604, + "grad_norm": 0.16313618421554565, + "kept_token_count": 737.0890625, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915118634700775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008488133270293474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.17718200683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.86217041015624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7670, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.0999406635761261, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016912457905709742, + "grad_norm": 0.11160529404878616, + "kept_token_count": 736.5796875, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9922889053821564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007711078971624374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.9165802001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.89374389648435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7680, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10072731748223304, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.015337951481342316, + "grad_norm": 0.13683408498764038, + "kept_token_count": 735.8375, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9913166582584381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008683346491307021, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.784033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.37627563476565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.10015259012579918, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.020859998371452092, + "grad_norm": 0.10163293033838272, + "kept_token_count": 736.1796875, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9908037126064301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009196308255195618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.74826965332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.4854034423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7700, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09994371309876442, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.01785298017784953, + "grad_norm": 0.13618366420269012, + "kept_token_count": 736.3265625, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.99088454246521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009115468710660934, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.13083801269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.7240417480469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7710, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09936991035938263, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.016174185555428267, + "grad_norm": 0.13470253348350525, + "kept_token_count": 737.0515625, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9911107301712037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008889251109212636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.79941711425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.8274261474609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7720, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10059901475906372, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.01860076477751136, + "grad_norm": 0.11762804538011551, + "kept_token_count": 735.9, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9917997717857361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008200239669531584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.27464904785154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.32012939453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7730, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.09943157136440277, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.015810808073729275, + "grad_norm": 0.125300794839859, + "kept_token_count": 737.296875, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9918197333812714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008180263731628657, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.02330017089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.26267395019534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7740, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.09964582547545434, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.01672693556174636, + "grad_norm": 0.12426076084375381, + "kept_token_count": 736.878125, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9916556239128113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0083443783223629, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.6937774658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.84214172363284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7750, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.1002371847629547, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.01673802100121975, + "grad_norm": 0.10885722190141678, + "kept_token_count": 736.1765625, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9910143494606019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008985626511275768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.28028869628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.11766052246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7760, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.09974664673209191, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.02201810209080577, + "grad_norm": 0.136480912566185, + "kept_token_count": 736.8859375, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9914248764514924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00857511917129159, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.6615509033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.2322692871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7770, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.0996446430683136, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.016490432154387235, + "grad_norm": 0.1020689606666565, + "kept_token_count": 736.909375, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9907440781593323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009255928639322519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1113647460937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.86493225097655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.09953937083482742, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016089604049921036, + "grad_norm": 0.13337451219558716, + "kept_token_count": 736.675, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9912885785102844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008711413107812404, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4534393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.78931579589846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7790, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.09921865612268448, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.015311253629624844, + "grad_norm": 0.1399175226688385, + "kept_token_count": 736.965625, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9914888322353363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00851117428392172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.24547119140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.4365631103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7800, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.10015667900443077, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.0172318272292614, + "grad_norm": 0.12130552530288696, + "kept_token_count": 736.440625, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9905794143676758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009420587960630656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.3082305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.4239166259766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7810, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.10011133998632431, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.014724796637892723, + "grad_norm": 0.11290057748556137, + "kept_token_count": 736.4703125, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9917094767093658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00829051248729229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.8258514404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.25954284667966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7820, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.09957734942436218, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.018713164422661067, + "grad_norm": 0.1182885691523552, + "kept_token_count": 736.875, + "learning_rate": 2.649527093199434e-05, + "loss": 0.023, + "mgd_cos_sim": 0.991351318359375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008648683503270149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.0147338867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.25059814453124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7830, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.10020807310938835, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.0165315393358469, + "grad_norm": 0.12461031973361969, + "kept_token_count": 736.6078125, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9918840646743774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008115934673696756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.7252136230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.7477691650391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7840, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.09973447322845459, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.01595655381679535, + "grad_norm": 0.15259742736816406, + "kept_token_count": 736.6375, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9913909435272217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008609058149158955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.677197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.9171142578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7850, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.09996909126639367, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.017368569783866406, + "grad_norm": 0.10892195999622345, + "kept_token_count": 736.446875, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.991878867149353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008121148683130741, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.2529754638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.11338806152344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.0997090108692646, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.01571836331859231, + "grad_norm": 0.13580115139484406, + "kept_token_count": 736.55, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9911274194717408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008872568514198065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.45786437988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.9485778808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7870, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10024595856666565, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.017204201221466063, + "grad_norm": 0.10980948805809021, + "kept_token_count": 736.13125, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9918898582458496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008110142778605223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.9712707519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.4710296630859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7880, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.10005573481321335, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.01796814436092973, + "grad_norm": 0.11531943827867508, + "kept_token_count": 736.215625, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9913345694541931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008665431011468171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.29200439453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.1686645507813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7890, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.10018884465098381, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.01769700851291418, + "grad_norm": 0.14254039525985718, + "kept_token_count": 736.1421875, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9910395801067352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008960416074842215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.3698028564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.9239807128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7900, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.10027193129062653, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.01726561915129423, + "grad_norm": 0.10407692939043045, + "kept_token_count": 736.2203125, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920351803302765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00796480719000101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.8571075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.881201171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7910, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.09951998069882392, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.016205355990678072, + "grad_norm": 0.08634507656097412, + "kept_token_count": 736.76875, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920985639095307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007901440560817718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1401672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.53795166015624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7920, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.10017666667699814, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.017454745154827832, + "grad_norm": 0.12080098688602448, + "kept_token_count": 736.3640625, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9917636692523957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008236326649785042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.7205352783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.31590270996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7930, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.10060612484812737, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.0176042802631855, + "grad_norm": 0.1389698088169098, + "kept_token_count": 735.7234375, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9911752879619599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008824714738875628, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.15451965332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.0884979248047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7940, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.09990176707506179, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.017902581579983236, + "grad_norm": 0.13637812435626984, + "kept_token_count": 736.4890625, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9914812564849853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008518735598772764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.25148315429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.65830078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.09980845972895622, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.01907819854095578, + "grad_norm": 0.11686964333057404, + "kept_token_count": 736.5171875, + "learning_rate": 2.635600617703457e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9921815395355225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007818456087261439, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.75755920410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.31280212402345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7960, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.0997658796608448, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.01964826313778758, + "grad_norm": 0.1076224222779274, + "kept_token_count": 736.5828125, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914017081260681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00859828144311905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1553649902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.4484405517578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10078424885869026, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.01708070794120431, + "grad_norm": 0.13490280508995056, + "kept_token_count": 735.6328125, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919041335582733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008095883205533028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.3639678955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.4151947021484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7980, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.09979600235819816, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.018883072212338446, + "grad_norm": 0.12026794254779816, + "kept_token_count": 736.65625, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9908657431602478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009134250320494175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.02630920410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.9319549560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 7990, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.10036716684699058, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.017148985620588066, + "grad_norm": 0.24078625440597534, + "kept_token_count": 736.146875, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9905694603919983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009430539421737194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.3002197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.464013671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8000, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10023863688111305, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.015496048890054227, + "grad_norm": 0.17068608105182648, + "kept_token_count": 736.2734375, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9913307726383209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008669234532862901, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.16734313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.23980407714845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8010, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.10022462233901024, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.016314833983778955, + "grad_norm": 0.12491470575332642, + "kept_token_count": 736.3515625, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9917057991027832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294199872761965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.46539306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.5900146484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8020, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09991691261529922, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.017188234720379114, + "grad_norm": 0.11170176416635513, + "kept_token_count": 736.6375, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9913371324539184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00866286689415574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.8796356201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.2976470947266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8030, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.09950466677546502, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.017777605447918177, + "grad_norm": 0.14423657953739166, + "kept_token_count": 737.203125, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.991398811340332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00860118055716157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.2286773681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.78325805664065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8040, + "valid_token_count": 818.665625 + }, + { + "actual_token_mask_ratio": 0.09993365779519081, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.017725744098424912, + "grad_norm": 0.1497763991355896, + "kept_token_count": 736.596875, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9911605417728424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0088394733145833, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.5524200439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.5374725341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8050, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.0999930314719677, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.017400129605084656, + "grad_norm": 0.12866170704364777, + "kept_token_count": 736.5078125, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915070474147797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008492943178862334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.86317443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.0394805908203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8060, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10078263953328133, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.016508649196475746, + "grad_norm": 0.1325782686471939, + "kept_token_count": 735.653125, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9915736675262451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008426334336400032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.1214538574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.7784423828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8070, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10012639909982682, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.018194449692964555, + "grad_norm": 0.1205611377954483, + "kept_token_count": 736.434375, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9917175650596619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008282439224421978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.52329711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.85150756835935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8080, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.10015316158533097, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.01599973253905773, + "grad_norm": 0.12593446671962738, + "kept_token_count": 736.3234375, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918546617031098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008145328145474195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.99107055664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.4552856445313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8090, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.09998863339424133, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.01984247723594308, + "grad_norm": 0.10943784564733505, + "kept_token_count": 736.328125, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9905274450778961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009472570195794106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.32325134277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.39947204589845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8100, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.10056740120053291, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.01693374030292034, + "grad_norm": 0.14503464102745056, + "kept_token_count": 736.1203125, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9911841571331024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008815842587500811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.62584533691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.0072937011719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8110, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.09962312579154968, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.018906142096966505, + "grad_norm": 0.13356497883796692, + "kept_token_count": 736.396875, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9911485433578491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008851455058902502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.0843780517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.521923828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8120, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.09935867190361022, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.018294112291187048, + "grad_norm": 0.12547270953655243, + "kept_token_count": 736.825, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9916825890541077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008317411597818136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.01036376953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.02277221679685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8130, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09949651956558228, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.015528038516640663, + "grad_norm": 0.09282911568880081, + "kept_token_count": 736.7375, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9913709819316864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008629013132303954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6420074462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.59690246582034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8140, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.10065284594893456, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.01854765843600035, + "grad_norm": 0.1250447928905487, + "kept_token_count": 735.8625, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9919286131858825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008071393705904484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9752471923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.5590423583984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8150, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09996016323566437, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.01809104923158884, + "grad_norm": 0.11109919100999832, + "kept_token_count": 736.5890625, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9911236345767975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008876371569931507, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.6671630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.38486633300784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8160, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10019613280892373, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.016868937574326993, + "grad_norm": 0.09743195027112961, + "kept_token_count": 736.5609375, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9913377344608307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008662260696291923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.8278564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.038525390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8170, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.09969145059585571, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.01576838120818138, + "grad_norm": 0.11204826831817627, + "kept_token_count": 736.3265625, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9918971240520478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008102879114449024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.1224792480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.5767791748047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8180, + "valid_token_count": 817.8578125 + }, + { + "actual_token_mask_ratio": 0.09979770705103874, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.01958972718566656, + "grad_norm": 0.11525942385196686, + "kept_token_count": 736.625, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9908656597137451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009134346805512904, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.5593200683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.00103454589845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8190, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.0999993346631527, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.016392682678997517, + "grad_norm": 0.10276094824075699, + "kept_token_count": 736.375, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9912554979324341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008744522742927075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.981982421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.00535583496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.1004735603928566, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.01710621677339077, + "grad_norm": 0.12890134751796722, + "kept_token_count": 735.909375, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9911527395248413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008847266342490912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.3511047363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.9908203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8210, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.10040989220142364, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.016651931591331957, + "grad_norm": 0.11501555144786835, + "kept_token_count": 736.1640625, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9913328230381012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008667187020182609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0588897705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.2359954833984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09990395084023476, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.01632838845252991, + "grad_norm": 0.12492477893829346, + "kept_token_count": 736.709375, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9913833022117615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00861669033765793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.44048767089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.96851501464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8230, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.10061827003955841, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.016733841225504874, + "grad_norm": 0.09839586168527603, + "kept_token_count": 736.040625, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9916473448276519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00835266588255763, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.9668426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.9444152832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8240, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.09992541149258613, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.01765090776607394, + "grad_norm": 0.12022511661052704, + "kept_token_count": 736.6, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920448184013366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007955169025808572, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.9336212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.1586212158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8250, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.09897727742791176, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.016546628344804047, + "grad_norm": 0.15783946216106415, + "kept_token_count": 736.8703125, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9913023352622986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008697670698165894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.9931182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.1637390136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8260, + "valid_token_count": 817.8125 + }, + { + "actual_token_mask_ratio": 0.09958949983119965, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.016462685260921716, + "grad_norm": 0.11070309579372406, + "kept_token_count": 736.9140625, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9921096801757813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007890322897583246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.58022155761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.01349182128905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8270, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.09993063285946846, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.018339811265468596, + "grad_norm": 0.07915368676185608, + "kept_token_count": 736.6484375, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9919107496738434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00808925237506628, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.39072570800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.74847412109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8280, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.09974581524729728, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.018207389395684004, + "grad_norm": 0.18215547502040863, + "kept_token_count": 737.0015625, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9914290070533752, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008570987917482852, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.83799438476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.38714599609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8290, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.10066546201705932, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.016943725943565368, + "grad_norm": 0.12703479826450348, + "kept_token_count": 735.93125, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9915119647979737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008488040510565042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.16298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.21153564453124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8300, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.1001775547862053, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.01857647215947509, + "grad_norm": 0.12745386362075806, + "kept_token_count": 736.39375, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9916980624198913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00830193879082799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.6794189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.8412353515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8310, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.10041655972599983, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.016944901645183565, + "grad_norm": 0.1459362804889679, + "kept_token_count": 736.034375, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9908512949943542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00914870873093605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.0099822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.769189453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8320, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.10008322075009346, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.017535741720348597, + "grad_norm": 0.11044895648956299, + "kept_token_count": 736.0765625, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9915944933891296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008405505679547787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.87209167480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.53826293945315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8330, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.09986416026949882, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.01667668707668781, + "grad_norm": 0.11066357046365738, + "kept_token_count": 736.434375, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.991037106513977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008962899073958397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.7157348632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.64439392089844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8340, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.09990884214639664, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.019302935618907214, + "grad_norm": 0.15017347037792206, + "kept_token_count": 736.3578125, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9910383522510529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008961641322821378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.13072204589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.1579864501953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8350, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.10017258003354072, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.016861644852906466, + "grad_norm": 0.16480658948421478, + "kept_token_count": 736.3359375, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9915322721004486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008467724826186896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.8776336669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.56159057617185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8360, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.09952875524759293, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.014599915128201246, + "grad_norm": 0.0870376005768776, + "kept_token_count": 736.546875, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9915268659591675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0084731362760067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.6015167236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.86663513183595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8370, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.1004704512655735, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.017118969466537237, + "grad_norm": 0.11447206884622574, + "kept_token_count": 736.0296875, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9915189206600189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008481071796268225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.3196624755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.23431396484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8380, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09963769167661667, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.016570748295634985, + "grad_norm": 0.09649407118558884, + "kept_token_count": 736.60625, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.991998428106308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008001561649143696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.3701904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.48082275390624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8390, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.10036730021238327, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.01613291408866644, + "grad_norm": 0.10434626787900925, + "kept_token_count": 736.05625, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920441210269928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007955885864794255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.90704040527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.0654754638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8400, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.1002182811498642, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.018125971779227257, + "grad_norm": 0.11233657598495483, + "kept_token_count": 736.2640625, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9918660521507263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008133953716605902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.65801696777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.5976806640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8410, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09986184686422347, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.01694326540455222, + "grad_norm": 0.14648471772670746, + "kept_token_count": 736.49375, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9917715430259705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008228451758623124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.6212158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.7781921386719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8420, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.09981335625052452, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.01522832727059722, + "grad_norm": 0.09337688982486725, + "kept_token_count": 736.765625, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9917253196239472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008274665288627147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.160791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.0121765136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8430, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.1001124456524849, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.016707710083574058, + "grad_norm": 0.1157337874174118, + "kept_token_count": 736.4, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9916617214679718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008338287379592657, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.8758544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.0299774169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8440, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.10065965801477432, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.020255328994244336, + "grad_norm": 0.13051162660121918, + "kept_token_count": 736.153125, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9919157922267914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008084225840866566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.3724822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.3155456542969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8450, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.09961238950490951, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.016849004477262498, + "grad_norm": 0.12253305315971375, + "kept_token_count": 736.8203125, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.991506141424179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008493846002966165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.8854156494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.3877319335937, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8460, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10041424483060837, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.01754179010167718, + "grad_norm": 0.1275222897529602, + "kept_token_count": 735.7875, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9917659223079681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008234083838760852, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.42588500976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.200634765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8470, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.1001813143491745, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.015996510069817303, + "grad_norm": 0.12496126443147659, + "kept_token_count": 736.3671875, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918206334114075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008179353922605515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.2025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.8227935791016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8480, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.09988135620951652, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.016855196189135314, + "grad_norm": 0.1381555199623108, + "kept_token_count": 736.3390625, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9920198976993561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00798010928556323, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.5819152832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.60537109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8490, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.10095311999320984, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.017931036185473204, + "grad_norm": 0.1426704078912735, + "kept_token_count": 735.5015625, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9914256930351257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008574316278100014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.49754028320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.9978454589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8500, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.1000349059700966, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.01633866475895047, + "grad_norm": 0.12420962005853653, + "kept_token_count": 736.31875, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9920242011547089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007975784596055746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.5664764404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.8233123779297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8510, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.10053266286849975, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.016118703223764897, + "grad_norm": 0.1127757877111435, + "kept_token_count": 736.09375, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9914131164550781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008586884941905737, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.70781860351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.5946319580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8520, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09995439797639846, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.016881019808351994, + "grad_norm": 0.16068492829799652, + "kept_token_count": 736.590625, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917820930480957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008217930980026721, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.61261291503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.4482940673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8530, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10078354924917221, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.017942771129310132, + "grad_norm": 0.12936246395111084, + "kept_token_count": 735.534375, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9920450866222381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007954928278923034, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.13197937011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.95042114257814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8540, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.10018565803766251, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.015370054356753825, + "grad_norm": 0.12819665670394897, + "kept_token_count": 736.4546875, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.991976922750473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008023067004978656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7153259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.51800537109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8550, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.09999609142541885, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.01909653227776289, + "grad_norm": 0.13879279792308807, + "kept_token_count": 736.378125, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9914341688156127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00856583472341299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.72158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.2806884765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8560, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.09964118972420692, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.02085668798536062, + "grad_norm": 0.09512791037559509, + "kept_token_count": 736.834375, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9912892937660217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008710696641355754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2461212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.6546356201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8570, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.10049650222063064, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.017986029852181673, + "grad_norm": 0.14688648283481598, + "kept_token_count": 736.234375, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9917118191719055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008288181759417057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.1388244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.1885284423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8580, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.10061315968632698, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.016650142800062894, + "grad_norm": 0.15645025670528412, + "kept_token_count": 736.00625, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9916815340518952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008318454213440418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.2740783691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.77763671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09971821457147598, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.015596960764378309, + "grad_norm": 0.12482330948114395, + "kept_token_count": 736.8890625, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9914818525314331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00851814802736044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2222839355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8014739990234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8600, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.09928648844361306, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.018291202653199435, + "grad_norm": 0.11950133740901947, + "kept_token_count": 737.3328125, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9915799021720886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00842008851468563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.9093078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.9627258300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8610, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.09972679540514946, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.016684248670935632, + "grad_norm": 0.10790786147117615, + "kept_token_count": 736.6671875, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9911114573478699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008888538833707571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.0229797363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.6933319091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8620, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10067587569355965, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.01815020004287362, + "grad_norm": 0.1348452866077423, + "kept_token_count": 735.9078125, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9922441244125366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007755865715444088, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.22198181152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.79771423339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8630, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09988666325807571, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.01849159160628915, + "grad_norm": 0.14667651057243347, + "kept_token_count": 736.575, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9915229976177216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008476994559168815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.93267517089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8595947265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8640, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.09971268102526665, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.0172903704456985, + "grad_norm": 0.11790110915899277, + "kept_token_count": 736.56875, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923583626747131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007641649153083563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.53612060546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.97258911132815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.10023574605584144, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.01610990767367184, + "grad_norm": 0.12077801674604416, + "kept_token_count": 736.196875, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9919691205024719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008030868135392666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.09717712402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.6361358642578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8660, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10004858151078225, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.018651621602475642, + "grad_norm": 0.12603603303432465, + "kept_token_count": 736.24375, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9913568437099457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00864316513761878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.51304931640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.43470458984376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8670, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.10014241859316826, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.018435462098568677, + "grad_norm": 0.13173113763332367, + "kept_token_count": 736.246875, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9915879130363464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008412064425647259, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4469024658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.765185546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8680, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.10004056319594383, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.01567027410492301, + "grad_norm": 0.1198495402932167, + "kept_token_count": 736.1765625, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9917920112609864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00820797374472022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.33363342285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.0857696533203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8690, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.1003181666135788, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.01479517612606287, + "grad_norm": 0.10988378524780273, + "kept_token_count": 736.24375, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9919237494468689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008076260704547167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.9695251464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.0264129638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.10064337998628617, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.019141959119588138, + "grad_norm": 0.1730796992778778, + "kept_token_count": 735.7796875, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.991327953338623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008672049548476935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.6702087402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.5198547363281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8710, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.10018330588936805, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.01673806132748723, + "grad_norm": 0.12214438617229462, + "kept_token_count": 736.4765625, + "learning_rate": 2.549613140294764e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9914986491203308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008501364383846521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.781982421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.8301971435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8720, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.10063271895051003, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.016197397839277983, + "grad_norm": 0.14628496766090393, + "kept_token_count": 736.3375, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9915900886058807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008409896306693554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.47997131347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.2813323974609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8730, + "valid_token_count": 818.7296875 + }, + { + "actual_token_mask_ratio": 0.10076895952224732, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.0166376743465662, + "grad_norm": 0.12568511068820953, + "kept_token_count": 735.8390625, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9921052038669587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007894802466034889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.0009338378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.0786895751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8740, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09984258860349655, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.015177713241428137, + "grad_norm": 0.14295129477977753, + "kept_token_count": 736.4265625, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9911900579929351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008809940703213216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0922882080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.6154479980469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8750, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.10015862286090851, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.015635014977306128, + "grad_norm": 0.1136971265077591, + "kept_token_count": 736.55625, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921930491924286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007806940469890833, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.9584686279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.7388397216797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8760, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.09963271170854568, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.01911760289222002, + "grad_norm": 0.1100880429148674, + "kept_token_count": 736.5765625, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9920048713684082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007995142601430416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.43004455566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.709228515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8770, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.09978105574846267, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.017749984934926032, + "grad_norm": 0.14564423263072968, + "kept_token_count": 736.7765625, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9917656004428863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008234399557113647, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.46347961425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.4963012695313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8780, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.09939854890108109, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.016964040417224168, + "grad_norm": 0.14441345632076263, + "kept_token_count": 736.9109375, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.991499000787735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008501006104052067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.2343994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7129760742188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8790, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.10036395341157914, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.013755805045366287, + "grad_norm": 0.13557805120944977, + "kept_token_count": 736.2203125, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.991663247346878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008336745947599412, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4064880371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.69142761230466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09951708763837815, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.015073462016880512, + "grad_norm": 0.10586647689342499, + "kept_token_count": 736.9578125, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9912367820739746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008763213269412517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.53284912109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3676330566406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8810, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.09971430376172066, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.01663088034838438, + "grad_norm": 0.11569063365459442, + "kept_token_count": 736.746875, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9920784533023834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007921552937477826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.7506622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.8798126220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8820, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.09989112913608551, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.01802310748025775, + "grad_norm": 0.114286869764328, + "kept_token_count": 736.678125, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915185153484345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008481493592262268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.9237762451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.6385223388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09953600019216538, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.019333165884017945, + "grad_norm": 0.10714726150035858, + "kept_token_count": 736.815625, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918830931186676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008116898033767939, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.4741973876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.4258972167969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8840, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09947662800550461, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.018728714995086194, + "grad_norm": 0.11397085338830948, + "kept_token_count": 736.8265625, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918794453144073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00812056614086032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.25025634765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.3576965332031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8850, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10011286735534668, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.017073577176779507, + "grad_norm": 0.12476742267608643, + "kept_token_count": 736.3734375, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919720947742462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008027899451553822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.6473785400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.57468872070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8860, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10018678829073906, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.018750435393303632, + "grad_norm": 0.13213977217674255, + "kept_token_count": 736.2484375, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9912746012210846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00872539086267352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.38999633789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.3370666503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8870, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.10050268769264221, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.015507004875689744, + "grad_norm": 0.11591395735740662, + "kept_token_count": 736.121875, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9914845943450927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008515393547713756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0251922607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.26539306640626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.10046123713254929, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.017968215979635715, + "grad_norm": 0.10969621688127518, + "kept_token_count": 736.146875, + "learning_rate": 2.529349093499642e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9918968915939331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008103119675070047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.07432861328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.40663757324216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8890, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.09997660219669342, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.016880830377340318, + "grad_norm": 0.10147639364004135, + "kept_token_count": 736.484375, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9915976703166962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008402307704091071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.06043395996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.742431640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8900, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.10001694709062577, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.01565362773835659, + "grad_norm": 0.12994372844696045, + "kept_token_count": 736.5484375, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9915419757366181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008458024263381958, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.0693389892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.29180297851565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.1005488820374012, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.016976659744977952, + "grad_norm": 0.11907050013542175, + "kept_token_count": 735.7953125, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9915091514587402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008490838203579187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.33453369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.2323303222656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8920, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.10000495836138726, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.02158753704279661, + "grad_norm": 0.11423661559820175, + "kept_token_count": 736.3859375, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9911889553070068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008811033330857754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8378082275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8697479248047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8930, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.10050172805786133, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.01889057094231248, + "grad_norm": 0.14088404178619385, + "kept_token_count": 736.2265625, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916418254375458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008358168788254261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.0608337402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.1363922119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8940, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.10000242590904236, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.016020337864756585, + "grad_norm": 0.10354897379875183, + "kept_token_count": 736.7171875, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918278515338897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008172154612839221, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.17479553222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.40282287597654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8950, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.09974921867251396, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.018585180025547744, + "grad_norm": 0.11657135933637619, + "kept_token_count": 736.5234375, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9919174790382386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008082502894103527, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5540283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.12493591308595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8960, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.0998920239508152, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.019320459757000207, + "grad_norm": 0.13147349655628204, + "kept_token_count": 736.4015625, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918303549289703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008169630449265242, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.35777282714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8970, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.09943763241171837, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.017046753037720918, + "grad_norm": 0.1339254230260849, + "kept_token_count": 736.75625, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9916324138641357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008367590513080359, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.5298309326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.94321899414064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8980, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.09982215613126755, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.017872101068496703, + "grad_norm": 0.11608430743217468, + "kept_token_count": 736.678125, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9917977690696717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008202222362160682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7093078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.9487701416016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 8990, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.10020180642604828, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.01708052624017, + "grad_norm": 0.11456400901079178, + "kept_token_count": 736.19375, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.991977471113205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008022525534033775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.53515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.8171356201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9000, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.0997882328927517, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.01822959827259183, + "grad_norm": 0.15998569130897522, + "kept_token_count": 736.8953125, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9916864812374115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008313524723052978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.310498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.92327880859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9010, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.10026177689433098, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.01848316229879856, + "grad_norm": 0.14765340089797974, + "kept_token_count": 736.1890625, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.991251140832901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00874884780496359, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.70018005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8294342041016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9020, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.09989760667085648, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.018201325833797456, + "grad_norm": 0.14384756982326508, + "kept_token_count": 736.5828125, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9912331819534301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00876681562513113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.8022430419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.60167541503904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9030, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.09943144917488098, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.016242800280451775, + "grad_norm": 0.11484270542860031, + "kept_token_count": 736.86875, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913937985897064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008606197871267795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.17047729492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.86399841308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9040, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.10036349967122078, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.015391883440315723, + "grad_norm": 0.10337059199810028, + "kept_token_count": 736.48125, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.99157355427742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008426441811025143, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.13233642578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.4099395751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9050, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.09981312826275826, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.01738497372716665, + "grad_norm": 0.12296965718269348, + "kept_token_count": 736.8671875, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9917886793613434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008211293909698724, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.69712524414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.41802978515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9060, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1000410832464695, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.018255373183637857, + "grad_norm": 0.11990579217672348, + "kept_token_count": 736.4046875, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9912043750286103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008795627672225238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.24078979492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.2682800292969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9070, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.09969153478741646, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.016608432121574878, + "grad_norm": 0.10839564353227615, + "kept_token_count": 737.0265625, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9916129946708679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008387000672519207, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2448974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.4230651855469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9080, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.09997159987688065, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.016536579560488464, + "grad_norm": 0.14580997824668884, + "kept_token_count": 736.646875, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9914771020412445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852290354669094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.5121185302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.68019409179686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9090, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.0997837170958519, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.01740780510008335, + "grad_norm": 0.11125971376895905, + "kept_token_count": 736.646875, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9917763769626617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008223626110702754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.68187255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.6082244873047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9100, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.09975714534521103, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.015474688354879617, + "grad_norm": 0.1514885425567627, + "kept_token_count": 736.6984375, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9912750959396363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008724897354841232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.95299377441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.8016632080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09967095106840133, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.017135797161608936, + "grad_norm": 0.10974239557981491, + "kept_token_count": 736.696875, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9916376054286957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008362390752881765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.7309875488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.2803985595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9120, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09986410439014434, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.016007322818040848, + "grad_norm": 0.1038626879453659, + "kept_token_count": 736.6890625, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.992449426651001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007550579309463501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.76727905273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.8448791503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9130, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.09968334510922432, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.015327052585780621, + "grad_norm": 0.09096340090036392, + "kept_token_count": 736.80625, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9916336953639984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00836630053818226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.06361999511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.61993408203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9140, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.09950210899114609, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.019507197383791208, + "grad_norm": 0.13659444451332092, + "kept_token_count": 736.9328125, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9914407849311828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008559228014200925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.3476837158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7699859619141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9150, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1000180907547474, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.015316730085760354, + "grad_norm": 0.09826183319091797, + "kept_token_count": 736.3796875, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9916124165058136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00838758870959282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.87877502441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7276123046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9160, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09979355186223984, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.01749365171417594, + "grad_norm": 0.12456974387168884, + "kept_token_count": 736.58125, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9915225386619568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008477450348436832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.29913635253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.3197784423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9170, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.1004999302327633, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.016315786819905042, + "grad_norm": 0.09281953424215317, + "kept_token_count": 735.7703125, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9909520208835602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009047973062843084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3936706542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7351470947266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9180, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.09990051239728928, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.018182992935180664, + "grad_norm": 0.18523451685905457, + "kept_token_count": 736.6734375, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9915508091449737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008449194021522999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.59930419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8259002685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9190, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09958973303437232, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.017271007969975473, + "grad_norm": 0.12191163003444672, + "kept_token_count": 736.6921875, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9919735252857208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008026477042585611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.36353454589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.70416564941405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09936182126402855, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.01825909670442343, + "grad_norm": 0.1434248387813568, + "kept_token_count": 736.80625, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.991848248243332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008151746168732644, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.89337158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.39508361816405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9210, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.09992282092571259, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.01752882618457079, + "grad_norm": 0.1487949788570404, + "kept_token_count": 736.4359375, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9914207339286805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008579267747700214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.73150939941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.7828094482422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9220, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10044794082641602, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.017600813508033754, + "grad_norm": 0.12530852854251862, + "kept_token_count": 736.2484375, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9913161158561706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008683890383690596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.34349670410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.77870788574216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9230, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.10018435716629029, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.01579801794141531, + "grad_norm": 0.12274643033742905, + "kept_token_count": 736.365625, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9911752939224243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008824715577065944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8245361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.2589904785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9240, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10025251358747482, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.017859815061092375, + "grad_norm": 0.12959463894367218, + "kept_token_count": 736.3234375, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9921228647232055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007877136766910552, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.00794677734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3668487548828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9250, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.09963639080524445, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.017996040917932988, + "grad_norm": 0.12639591097831726, + "kept_token_count": 736.6421875, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9918703615665436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008129642345011234, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.3218597412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3024627685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9260, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.1005834087729454, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.02010659147053957, + "grad_norm": 0.14525765180587769, + "kept_token_count": 736.409375, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9919468879699707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008053110167384147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.6890411376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.77142944335935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9270, + "valid_token_count": 818.765625 + }, + { + "actual_token_mask_ratio": 0.10001909881830215, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.01827617660164833, + "grad_norm": 0.1687210202217102, + "kept_token_count": 736.6421875, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9918485522270203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008151442743837833, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.9493408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.3471923828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9280, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.10033213347196579, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.015998197626322506, + "grad_norm": 0.13354641199111938, + "kept_token_count": 736.0734375, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9921048998832702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007895109616219998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.1752227783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.15118408203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9290, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.09937775954604149, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.01789605524390936, + "grad_norm": 0.12754300236701965, + "kept_token_count": 736.859375, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.992432051897049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007567958068102598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9294769287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.02358093261716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9300, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.1004242867231369, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.016512151155620814, + "grad_norm": 0.13665948808193207, + "kept_token_count": 735.796875, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9918971121311188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008102887962013482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.0991607666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6544250488281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9310, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.10020414963364602, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.015162497758865356, + "grad_norm": 0.11161680519580841, + "kept_token_count": 736.4828125, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9916752994060516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008324694819748402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.48968811035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.3185668945313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9320, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10015647932887077, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.016660323925316335, + "grad_norm": 0.1536584198474884, + "kept_token_count": 736.3546875, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9915430665016174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008456940948963165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.65586242675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.3094970703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9330, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.09924140870571137, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.021284627355635168, + "grad_norm": 0.118316151201725, + "kept_token_count": 737.1265625, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9918947994709015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008105201087892056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1322509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.1568267822266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9340, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.10056239143013954, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.017520935367792846, + "grad_norm": 0.10222097486257553, + "kept_token_count": 736.03125, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9916785478591919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008321459591388702, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.7575958251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.80186767578124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.10034498944878578, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.01660895477980375, + "grad_norm": 0.14657677710056305, + "kept_token_count": 736.25, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9918233156204224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008176691457629203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.8095642089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.54005126953126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10008477345108986, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.015592452418059111, + "grad_norm": 0.09481523931026459, + "kept_token_count": 736.3703125, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9915215015411377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008478492870926857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6718017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.00823669433595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10003548860549927, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014289842359721661, + "grad_norm": 0.08703798055648804, + "kept_token_count": 736.384375, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9919472575187683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00805273773148656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.5809326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.60174865722655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.1004350483417511, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.01668483726680279, + "grad_norm": 0.13923092186450958, + "kept_token_count": 736.2671875, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9913230299949646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008676969818770886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.74920959472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.89869995117186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9390, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.10057263150811195, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.017227482050657272, + "grad_norm": 0.1091628298163414, + "kept_token_count": 736.0171875, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9916070580482483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008392935432493687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.56298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.8303619384766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9400, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.1001489743590355, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.01824194611981511, + "grad_norm": 0.12620298564434052, + "kept_token_count": 736.5390625, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9913094282150269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008690585568547248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.0882171630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.70233154296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9410, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.1008326143026352, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015099512692540885, + "grad_norm": 0.10039936006069183, + "kept_token_count": 735.8015625, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9919586718082428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008041322790086269, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2219512939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 388.90540771484376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9420, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09967951774597168, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.015846847929060458, + "grad_norm": 0.11421564966440201, + "kept_token_count": 737.1140625, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9921758592128753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007824140973389149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.06878356933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.7021026611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9430, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.09998301491141319, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.016194137651473282, + "grad_norm": 0.13180264830589294, + "kept_token_count": 736.475, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9918428361415863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00815717587247491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.4861114501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.2423553466797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9440, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10033003687858581, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.017240140493959187, + "grad_norm": 0.17494022846221924, + "kept_token_count": 735.9203125, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9919666767120361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008033328596502543, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.68495178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.4853485107422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9450, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.10032386109232902, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.018416804634034632, + "grad_norm": 0.17085203528404236, + "kept_token_count": 736.1078125, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9906935930252075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009306390956044196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.5907867431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.65476684570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9460, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.10084452405571938, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.018049156852066518, + "grad_norm": 0.11282061040401459, + "kept_token_count": 735.9921875, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9914183855056763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008581619244068861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.9599212646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.8026824951172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9470, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.09991008937358856, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.01677080672234297, + "grad_norm": 0.1217091828584671, + "kept_token_count": 736.4125, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.991403067111969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008596920687705279, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.6684112548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.4222869873047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9480, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.09996526837348937, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.016563044022768737, + "grad_norm": 0.1162421926856041, + "kept_token_count": 736.309375, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919976770877839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00800232132896781, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.6158874511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4546813964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9490, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.1001360759139061, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.014375751838088036, + "grad_norm": 0.09923120588064194, + "kept_token_count": 736.3, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9918903529644012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008109640050679445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.4334991455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.81329040527345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9500, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10091848224401474, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.016740240436047316, + "grad_norm": 0.12786072492599487, + "kept_token_count": 735.7703125, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9922215342521667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00777846947312355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.98643493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8205627441406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9510, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.0996202290058136, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.018556355964392424, + "grad_norm": 0.11205289512872696, + "kept_token_count": 736.3578125, + "learning_rate": 2.451158484961621e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9911702573299408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008829729910939933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.5287658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.2063873291016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9520, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.09972158223390579, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.013897391594946384, + "grad_norm": 0.10609686374664307, + "kept_token_count": 736.66875, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9916730105876923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008326972648501397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.35130920410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.66106567382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9530, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.10057437568902969, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.01654631979763508, + "grad_norm": 0.0989316925406456, + "kept_token_count": 736.06875, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9917369365692139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008263058215379714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.5752471923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.4591033935547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9540, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.10011213198304177, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.016328411549329756, + "grad_norm": 0.11252367496490479, + "kept_token_count": 736.46875, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9919328451156616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008067161962389945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2049194335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.2301483154297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9550, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.09973349273204804, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.014407385047525168, + "grad_norm": 0.0952555313706398, + "kept_token_count": 736.7671875, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9922425329685212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007757470570504665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3257110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.21416625976565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9560, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.10004562959074974, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.016960938274860383, + "grad_norm": 0.10846588760614395, + "kept_token_count": 736.1921875, + "learning_rate": 2.444751372488677e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9920093059539795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00799069246277213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.2809631347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.62409057617185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9570, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.09964442476630211, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.017923984117805956, + "grad_norm": 0.12328788638114929, + "kept_token_count": 737.0734375, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9916187465190888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008381252735853195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0640563964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.3089263916016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9580, + "valid_token_count": 818.64375 + }, + { + "actual_token_mask_ratio": 0.10001052990555763, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.017574246879667042, + "grad_norm": 0.16593852639198303, + "kept_token_count": 736.34375, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9916810035705567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008319011982530355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.31345520019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.6859832763672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.09981048181653022, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017157492879778146, + "grad_norm": 0.14254489541053772, + "kept_token_count": 736.7390625, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9919020354747772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008097944036126136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.1994354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.2179656982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9600, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.10083644762635231, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.015363746788352728, + "grad_norm": 0.10747681558132172, + "kept_token_count": 735.565625, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9918690383434295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008130969759076834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0556701660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.9689056396484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9610, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.1001495100557804, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.01648693038150668, + "grad_norm": 0.10115238279104233, + "kept_token_count": 736.3015625, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9915572106838226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008442782703787089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.990869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8342712402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9620, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10053137242794037, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.01636315751820803, + "grad_norm": 0.10567349195480347, + "kept_token_count": 735.98125, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9920644044876099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00793559430167079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.91211853027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.5353698730469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9630, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.09939552694559098, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.016360856592655182, + "grad_norm": 0.11066490411758423, + "kept_token_count": 736.828125, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9919333577156066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008066642098128796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8444580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.35614013671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9640, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.1003297060728073, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.016190713830292226, + "grad_norm": 0.12346000969409943, + "kept_token_count": 736.3109375, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9918183147907257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008181683626025915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.01752319335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.84507751464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9650, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.10014980435371398, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.01503135273233056, + "grad_norm": 0.11932497471570969, + "kept_token_count": 736.59375, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9920803308486938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007919656578451395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.85365905761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.92064208984374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9660, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.10035206526517867, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.015514355525374413, + "grad_norm": 0.11265601217746735, + "kept_token_count": 736.2125, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9921389281749725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007861068286001683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.80364685058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.05674438476564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.09997537434101104, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015432230848819017, + "grad_norm": 0.09963098168373108, + "kept_token_count": 736.4796875, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924989819526673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007501034252345562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.5218505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.3353759765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9680, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09949875921010971, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.015781305078417063, + "grad_norm": 0.10456492006778717, + "kept_token_count": 736.690625, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9918978571891784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008102130983024836, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.1826995849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.24591369628905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9690, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.10038947314023972, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.01757977306842804, + "grad_norm": 0.11403243988752365, + "kept_token_count": 736.1484375, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9918322384357452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008167761471122503, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8430419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.6556091308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9700, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09979941695928574, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.01681954404339194, + "grad_norm": 0.10937722027301788, + "kept_token_count": 736.628125, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917189955711365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008281019516289234, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.7873870849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.9224548339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9710, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10079326555132866, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.016939023230224847, + "grad_norm": 0.11280952394008636, + "kept_token_count": 735.940625, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9920811295509339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791888264939189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.2939849853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.5746704101563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9720, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.09983278065919876, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.014004748780280352, + "grad_norm": 0.11016715317964554, + "kept_token_count": 736.6234375, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9921238243579864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0078761862590909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.90215759277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.34022521972656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9730, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.1008458212018013, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.018661281932145356, + "grad_norm": 0.14897632598876953, + "kept_token_count": 735.584375, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9916533291339874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00834666034206748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.6483673095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.2376739501953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9740, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.10039381608366967, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.017936095222830774, + "grad_norm": 0.12703846395015717, + "kept_token_count": 736.1359375, + "learning_rate": 2.421449545438209e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9918555736541748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008144439477473498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.6728057861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.8151336669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9750, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09925203919410705, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.01616414599120617, + "grad_norm": 0.12152691185474396, + "kept_token_count": 736.90625, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917292058467865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008270802535116672, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.8602294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.79212341308596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9760, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10007264167070389, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.015412337146699429, + "grad_norm": 0.07887256890535355, + "kept_token_count": 736.5984375, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9918295919895173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00817040791735053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.60703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9770, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.10048772692680359, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.018628438375890254, + "grad_norm": 0.1154417023062706, + "kept_token_count": 735.80625, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9918545782566071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008145411405712366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.62158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.5355712890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9780, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.1006019227206707, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.018872714787721633, + "grad_norm": 0.11216897517442703, + "kept_token_count": 736.059375, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9914649426937103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008535058330744505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.6347320556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.69492797851564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9790, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.09944169223308563, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.015202234033495188, + "grad_norm": 0.10165047645568848, + "kept_token_count": 737.1125, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9921884298324585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0078115747310221195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7044677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.94358215332034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9800, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.09977632313966751, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.014576206542551517, + "grad_norm": 0.10874519497156143, + "kept_token_count": 736.3578125, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9918737232685089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008126296754926442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.6686218261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8634765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9810, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.09972013533115387, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.015267257671803236, + "grad_norm": 0.10602773725986481, + "kept_token_count": 736.6640625, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9915027618408203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00849722595885396, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.58080139160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.18792114257815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9820, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.09978683367371559, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.01625218791887164, + "grad_norm": 0.14870481193065643, + "kept_token_count": 736.634375, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9918208301067353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008179156575351954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.12041625976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.9776336669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.0996799997985363, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.016592199075967075, + "grad_norm": 0.09387948364019394, + "kept_token_count": 736.5140625, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.99177485704422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008225151989609002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.4061767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.8062347412109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9840, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.09947110563516617, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.01752809602767229, + "grad_norm": 0.1701565980911255, + "kept_token_count": 736.4890625, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9920709192752838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007929087802767753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6995422363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.10935363769534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9850, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.09947068318724632, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.016044298466295005, + "grad_norm": 0.12085522711277008, + "kept_token_count": 736.60625, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917680501937867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008231948409229518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.83997497558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.5477783203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9860, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.10070156082510948, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.016811464447528124, + "grad_norm": 0.11483023315668106, + "kept_token_count": 736.3484375, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9922607958316803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007739190943539143, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.89051818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.1178924560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9870, + "valid_token_count": 818.803125 + }, + { + "actual_token_mask_ratio": 0.10021227076649666, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.01699050487950444, + "grad_norm": 0.13077162206172943, + "kept_token_count": 736.2640625, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9917419493198395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008258047699928283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7127258300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.3529296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9880, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.09956977143883705, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.017227052059024573, + "grad_norm": 0.1394871026277542, + "kept_token_count": 736.8296875, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9917208552360535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008279146626591682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.0644805908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.1879028320312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9890, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09997046738862991, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.016068623587489127, + "grad_norm": 0.12611174583435059, + "kept_token_count": 736.4109375, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9922688364982605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007731162384152413, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.4394989013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.8033813476562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9900, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.1000961683690548, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.01634276593104005, + "grad_norm": 0.14747890830039978, + "kept_token_count": 736.6109375, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9920243561267853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007975633163005114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.97479553222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.6575622558594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9910, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.100710941106081, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.019171486143022776, + "grad_norm": 0.11512822657823563, + "kept_token_count": 735.6765625, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9924399793148041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007560014352202415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.4598846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.0331512451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9920, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.0997018076479435, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.018961400259286165, + "grad_norm": 0.13143537938594818, + "kept_token_count": 736.7296875, + "learning_rate": 2.397784964272865e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9919855237007141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00801447294652462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3363555908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.50890197753904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9930, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.1001620464026928, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.016044005751609802, + "grad_norm": 0.11672823876142502, + "kept_token_count": 736.4609375, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9922156929969788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077842999249696735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.45147094726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.5673309326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9940, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.09903787449002266, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.015996125154197217, + "grad_norm": 0.1122908815741539, + "kept_token_count": 737.065625, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9921909689903259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007809044979512692, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.52228393554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.7843414306641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.10044693276286125, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.015295640006661415, + "grad_norm": 0.09220025688409805, + "kept_token_count": 735.9953125, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9916920363903046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00830796491354704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.6997314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.03160705566404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9960, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.09969742000102996, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.015923620387911796, + "grad_norm": 0.11550232768058777, + "kept_token_count": 736.915625, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9915619194507599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008438078686594963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.8078216552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.5919555664062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9970, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.10075236186385154, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.016286979150027036, + "grad_norm": 0.1094760000705719, + "kept_token_count": 736.1265625, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9916390836238861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008360904175788164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.319580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.0693359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9980, + "valid_token_count": 818.603125 + }, + { + "actual_token_mask_ratio": 0.10029537901282311, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.015186231583356857, + "grad_norm": 0.10543030500411987, + "kept_token_count": 736.20625, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9923212349414825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007678773067891598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.90220336914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9017700195312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 9990, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.10024712234735489, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.01636969894170761, + "grad_norm": 0.1169862151145935, + "kept_token_count": 736.109375, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9917741775512695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008225825428962708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1628845214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.0218933105469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10000, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10016327127814292, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.014837063290178776, + "grad_norm": 0.1216331273317337, + "kept_token_count": 736.534375, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9916816413402557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008318357449024916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.68223571777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.38397521972655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10010, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.1001763291656971, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.01707728561013937, + "grad_norm": 0.1247941330075264, + "kept_token_count": 736.3625, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9921329379081726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007867035735398532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.29112548828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8498596191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10020, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.10005272626876831, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.017154153250157834, + "grad_norm": 0.11923031508922577, + "kept_token_count": 736.7328125, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9914394915103912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008560506999492646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.87986450195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.64625244140626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10030, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.10002894625067711, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.01587338913232088, + "grad_norm": 0.1065841093659401, + "kept_token_count": 736.3234375, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9926439166069031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007356083393096924, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.9213592529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.18605346679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.09950032532215118, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.01472237091511488, + "grad_norm": 0.1099255383014679, + "kept_token_count": 736.721875, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9920822858810425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007917709648609161, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3701477050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.9378295898438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.10056393817067147, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.015136895701289177, + "grad_norm": 0.11694146692752838, + "kept_token_count": 736.053125, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9920493483543396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00795064577832818, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.18190307617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.5853759765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10060, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.0999617800116539, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.01583523824810982, + "grad_norm": 0.14319394528865814, + "kept_token_count": 736.534375, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9918341159820556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008165876381099224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.17279052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.3903564453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10070, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09947842210531235, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.01685837535187602, + "grad_norm": 0.12431243062019348, + "kept_token_count": 736.6375, + "learning_rate": 2.37779409311278e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917249560356141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00827503576874733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.86134643554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.90918884277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10080, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.09978415444493294, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.01573420986533165, + "grad_norm": 0.10375410318374634, + "kept_token_count": 736.5453125, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9917938709259033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008206139225512743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.4595611572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.95012817382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10090, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.09996510297060013, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.017604523804038763, + "grad_norm": 0.11096630990505219, + "kept_token_count": 736.59375, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9926447570323944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007355249766260385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.36830444335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.6666595458984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10056186690926552, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.017442192416638136, + "grad_norm": 0.13095806539058685, + "kept_token_count": 735.740625, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9913610458374024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00863896869122982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.8552551269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.2296203613281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10110, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.10023517683148384, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.016096153017133474, + "grad_norm": 0.17976902425289154, + "kept_token_count": 736.3390625, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922540605068206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007745947316288948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5179718017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.1261291503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10120, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.0998316265642643, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.014664657972753048, + "grad_norm": 0.11555901914834976, + "kept_token_count": 736.3421875, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9924358189105987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007564179413020611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7660308837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.7932830810547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10130, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.09977883398532868, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.016859098710119723, + "grad_norm": 0.15359263122081757, + "kept_token_count": 737.046875, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9916921019554138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008307906705886126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1866760253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 389.9383148193359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10140, + "valid_token_count": 818.7421875 + }, + { + "actual_token_mask_ratio": 0.10012748688459397, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.014162810146808624, + "grad_norm": 0.09587451815605164, + "kept_token_count": 736.38125, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.992116641998291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007883358746767044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.5824279785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.1716705322266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10150, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.10043296068906785, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.015099075809121131, + "grad_norm": 0.13085688650608063, + "kept_token_count": 736.025, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9913517713546753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00864822780713439, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.5153076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.0977355957031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10160, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.09941216185688972, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.018593458272516728, + "grad_norm": 0.11106926947832108, + "kept_token_count": 737.0640625, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.992009848356247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007990162633359433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.8406036376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.1433044433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10170, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.09989802464842797, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.015638381242752075, + "grad_norm": 0.12945891916751862, + "kept_token_count": 736.5046875, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9918521523475647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008147858176380396, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.6618347167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.27259521484376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10180, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1002967543900013, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.013568372000008821, + "grad_norm": 0.10861913114786148, + "kept_token_count": 736.171875, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9920800447463989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791995069012046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.6760559082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.9134460449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10190, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.10079958587884903, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.0157534871250391, + "grad_norm": 0.10869099199771881, + "kept_token_count": 736.053125, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9912627756595611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008737217169255018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.68711242675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.34747924804685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10200, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.0997778058052063, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.015428678598254919, + "grad_norm": 0.09152376651763916, + "kept_token_count": 736.7078125, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9919821083545685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008017888572067023, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.19810791015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.0727783203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10210, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.09952902793884277, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.017988767568022014, + "grad_norm": 0.11441274732351303, + "kept_token_count": 736.7953125, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9920875668525696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007912428118288517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.89515686035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.5057678222656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10220, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.0989946149289608, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.015497140027582646, + "grad_norm": 0.08772692084312439, + "kept_token_count": 737.015625, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920861959457398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007913806103169918, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.972802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.907861328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10230, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.099522764980793, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.015043978113681078, + "grad_norm": 0.08620811998844147, + "kept_token_count": 736.95, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9916792213916779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008320781867951155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.6483459472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.93572387695315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10056421533226967, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.014544004760682582, + "grad_norm": 0.12417902052402496, + "kept_token_count": 735.8796875, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9918588101863861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00814119353890419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.55858459472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.24989624023436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10250, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.10007379427552224, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.01713225245475769, + "grad_norm": 0.10524435341358185, + "kept_token_count": 736.3328125, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921281039714813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007871907763183117, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.2358154296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.1301208496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10260, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09999220967292785, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.01657870002090931, + "grad_norm": 0.12621738016605377, + "kept_token_count": 736.43125, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.992503696680069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007496311794966459, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.31817321777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.1520751953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10270, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.10041202530264855, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.016999041941016913, + "grad_norm": 0.11703614145517349, + "kept_token_count": 735.7921875, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9915373921394348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008462612051516771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6962646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.1058319091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10280, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.09955077320337295, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.015692710317671298, + "grad_norm": 0.1213330328464508, + "kept_token_count": 736.6578125, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9922519028186798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007748098578304052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.7703094482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.28891906738284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10290, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.0998145543038845, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.01683465102687478, + "grad_norm": 0.1111644059419632, + "kept_token_count": 736.6515625, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9919681251049042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008031873404979706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6114929199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.28822937011716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10300, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09980330243706703, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.0171476935967803, + "grad_norm": 0.13439211249351501, + "kept_token_count": 736.4453125, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9914964854717254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850351732224226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.3206329345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.17006225585936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10310, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.10016020461916923, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.01443697391077876, + "grad_norm": 0.11485238373279572, + "kept_token_count": 736.1578125, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.991452956199646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008547037560492753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.5076019287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.77539978027346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10320, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.09968013912439347, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.016946541238576175, + "grad_norm": 0.10915868729352951, + "kept_token_count": 736.83125, + "learning_rate": 2.343945413972411e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9918627977371216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008137195650488139, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.7756317138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 390.7961120605469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10330, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10065570920705795, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.015779077261686324, + "grad_norm": 0.13518886268138885, + "kept_token_count": 735.521875, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9913779437541962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00862205671146512, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.60736694335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.6242431640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10340, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.100215382874012, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.016147557366639375, + "grad_norm": 0.10333051532506943, + "kept_token_count": 736.1953125, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913108348846436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00868917554616928, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.2222564697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0993286132813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10350, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.10040097013115883, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.016862786188721657, + "grad_norm": 0.11091377586126328, + "kept_token_count": 736.3265625, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9924120962619781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007587901689112187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.5916473388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7351318359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10360, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.1000664085149765, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.016072407923638822, + "grad_norm": 0.10440139472484589, + "kept_token_count": 736.6734375, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922399818897247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007760006655007601, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.51995544433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.06463012695315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10370, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.10090147554874421, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.015486524254083634, + "grad_norm": 0.13970059156417847, + "kept_token_count": 735.7359375, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9914050877094269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008594914618879557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.87493286132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.6325958251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10380, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.09995927438139915, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.017754600383341312, + "grad_norm": 0.11946777254343033, + "kept_token_count": 736.478125, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9923227429389954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007677256874740124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.72898864746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.43896484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10390, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10044140592217446, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.01611320013180375, + "grad_norm": 0.10040691494941711, + "kept_token_count": 736.4703125, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920845210552216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007915476337075233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0879852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.2796997070312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10400, + "valid_token_count": 818.6984375 + }, + { + "actual_token_mask_ratio": 0.10078956186771393, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.016291778068989515, + "grad_norm": 0.09689953178167343, + "kept_token_count": 735.809375, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9917745530605316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008225446101278067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.06165771484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.03358459472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10410, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.09936965033411979, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.016185321472585203, + "grad_norm": 0.1381317377090454, + "kept_token_count": 736.7484375, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917959809303284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008204028382897377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.26764526367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.4489349365234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10420, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.10032035186886787, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014729375950992107, + "grad_norm": 0.10632681101560593, + "kept_token_count": 736.215625, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9919693350791932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008030680939555168, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.9318145751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.60155334472654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10430, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09915487170219421, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.015593099128454923, + "grad_norm": 0.13490179181098938, + "kept_token_count": 737.1703125, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920907616615295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007909242063760757, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.1306579589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.9359558105469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10440, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.09995530173182487, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.01632286552339792, + "grad_norm": 0.13478755950927734, + "kept_token_count": 736.3421875, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9922139883041382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007786011882126331, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5981781005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.70274353027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10450, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.0997110590338707, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.016939850337803364, + "grad_norm": 0.11425916850566864, + "kept_token_count": 736.46875, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9924476385116577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007552372571080923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.7514617919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.95335388183594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10460, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.10042665526270866, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015599837712943553, + "grad_norm": 0.13413774967193604, + "kept_token_count": 736.1, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9924443006515503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007555704656988382, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.36765441894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.55280151367185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10470, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.10078045055270195, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.015440566092729568, + "grad_norm": 0.13408879935741425, + "kept_token_count": 736.240625, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9917782247066498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008221786003559828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.88869018554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.94619140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10480, + "valid_token_count": 818.753125 + }, + { + "actual_token_mask_ratio": 0.10011483356356621, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.016743370797485113, + "grad_norm": 0.10599252581596375, + "kept_token_count": 736.284375, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9917908668518066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008209132775664329, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.0568115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.77618713378905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.10047140046954155, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.019629373401403426, + "grad_norm": 0.11830582469701767, + "kept_token_count": 736.06875, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9917284727096558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008271540142595768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.93007507324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.042578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10500, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.09971484914422035, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.014404891338199378, + "grad_norm": 0.10785581916570663, + "kept_token_count": 736.453125, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9920057356357574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007994273863732814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.3241821289063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.1744384765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10510, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.09982604905962944, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.015018756687641143, + "grad_norm": 0.12612241506576538, + "kept_token_count": 736.709375, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9921914756298065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007808514032512903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.69524841308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.14947204589845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10520, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.10076332241296768, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.018909234274178743, + "grad_norm": 0.13473917543888092, + "kept_token_count": 735.8515625, + "learning_rate": 2.316403786537544e-05, + "loss": 0.023, + "mgd_cos_sim": 0.99179105758667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008208927791565656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8774658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.5855682373047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10530, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09998916387557984, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.01695225602015853, + "grad_norm": 0.1223154067993164, + "kept_token_count": 736.2921875, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9920328974723815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00796709069982171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.00487060546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.4128753662109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10540, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.100106181204319, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.01817711126059294, + "grad_norm": 0.13504447042942047, + "kept_token_count": 736.0359375, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9914755284786224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008524464257061482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9061981201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.0858123779297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10550, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.09993372336030007, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.015233442839235067, + "grad_norm": 0.1140999123454094, + "kept_token_count": 736.3359375, + "learning_rate": 2.312237989864752e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9924593865871429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075406040996313095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0390167236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.62030029296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10560, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09983888864517212, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.016097963135689498, + "grad_norm": 0.10584034025669098, + "kept_token_count": 736.79375, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.99201420545578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007985783554613591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.02214965820315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.95567016601564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10570, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.09950789660215378, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.014623921271413565, + "grad_norm": 0.12098288536071777, + "kept_token_count": 736.534375, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9918806374073028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008119371440261602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.60695495605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.9103210449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10580, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.09976619631052017, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.015988509356975555, + "grad_norm": 0.12586525082588196, + "kept_token_count": 736.6421875, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9913398385047912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008660161029547453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2427581787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.59307250976565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10590, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.09963498562574387, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.01498631639406085, + "grad_norm": 0.10560650378465652, + "kept_token_count": 736.9828125, + "learning_rate": 2.306669784959729e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9919784188270568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008021572418510915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.8623840332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.4421905517578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10600, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.10011298358440399, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.014061572402715683, + "grad_norm": 0.1010015681385994, + "kept_token_count": 736.6875, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9924759447574616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007524037547409534, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5116912841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9923095703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10610, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.09926939085125923, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.017065592203289272, + "grad_norm": 0.11973267793655396, + "kept_token_count": 737.1265625, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9922213137149811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007778674736618996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.53948364257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5900054931641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10620, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.10052668675780296, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.016185083612799644, + "grad_norm": 0.10216084122657776, + "kept_token_count": 735.803125, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9925246894359588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007475311961024999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.0263305664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.12978515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10630, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.10015975385904312, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.018263372220098974, + "grad_norm": 0.1005672886967659, + "kept_token_count": 736.3765625, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9920339226722718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007966089993715286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.4427154541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.1731811523438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10640, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09996605440974235, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.0162692335434258, + "grad_norm": 0.10559356957674026, + "kept_token_count": 736.5453125, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9923247396945953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007675261329859495, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0696075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.23643493652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10650, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09969003349542618, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.012542632967233657, + "grad_norm": 0.10773955285549164, + "kept_token_count": 736.6, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9923595309257507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076404747553169726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.8639770507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.0479461669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10660, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.09971692338585854, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.015621643606573343, + "grad_norm": 0.0912160873413086, + "kept_token_count": 736.4578125, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9921340584754944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007865930255502463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.0553863525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4598724365234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10670, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.100257708132267, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.017010286264121533, + "grad_norm": 0.15482176840305328, + "kept_token_count": 736.125, + "learning_rate": 2.295486435147251e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9919636487960816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00803635260090232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.39073791503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6663818359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10680, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10062858536839485, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.016432210616767407, + "grad_norm": 0.11776290833950043, + "kept_token_count": 736.3375, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9915729284286499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00842707958072424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.70652160644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6089233398437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10690, + "valid_token_count": 818.728125 + }, + { + "actual_token_mask_ratio": 0.09925240129232407, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.016682924702763557, + "grad_norm": 0.09923043102025986, + "kept_token_count": 737.1953125, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9922491192817688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077508784830570224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.6049041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.1114990234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10700, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.09958239942789078, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.01624475298449397, + "grad_norm": 0.10852326452732086, + "kept_token_count": 736.6703125, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9926942646503448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073057269677519795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.8912719726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8114288330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10710, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.10081668421626092, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.01799778062850237, + "grad_norm": 0.11407431960105896, + "kept_token_count": 735.4578125, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9921584069728852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007841583900153637, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.544287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3062438964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10720, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.09960470423102379, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.016190076898783445, + "grad_norm": 0.1335948258638382, + "kept_token_count": 736.584375, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921370685100556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007862927205860615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0080627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2718505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10730, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.10064361691474914, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.017215682566165923, + "grad_norm": 0.10701539367437363, + "kept_token_count": 735.8421875, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923043370246887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00769566697999835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9956878662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.21265869140626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10740, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.0997196950018406, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.01664457144215703, + "grad_norm": 0.13434478640556335, + "kept_token_count": 736.7203125, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9923058450222015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007694156281650067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.7416564941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4399047851563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10750, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.09971499517560005, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.016262699849903584, + "grad_norm": 0.10772000998258591, + "kept_token_count": 736.5390625, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9916693687438964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008330628927797078, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.6504150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2689605712891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10760, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.10036160424351692, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.015461435355246068, + "grad_norm": 0.1120242327451706, + "kept_token_count": 735.853125, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9922403991222382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007759602461010218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8724731445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7414184570313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10770, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.10056676492094993, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.013434016378596425, + "grad_norm": 0.07249260693788528, + "kept_token_count": 736.1625, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9921395778656006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00786041161045432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.25511169433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8047760009766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.09982755035161972, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.014682883862406015, + "grad_norm": 0.1549849510192871, + "kept_token_count": 736.5203125, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9923729419708252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007627058029174805, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.0585052490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.5807800292969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10790, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.10015647113323212, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.01817335020750761, + "grad_norm": 0.172111377120018, + "kept_token_count": 736.378125, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9912232100963593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008776797726750375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6854614257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1634552001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09998138025403022, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.015669273119419812, + "grad_norm": 0.10601732134819031, + "kept_token_count": 736.6015625, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9926185131072998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007381498347967863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.6408447265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.13344116210936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10810, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.10021277368068696, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.015884893480688334, + "grad_norm": 0.11767059564590454, + "kept_token_count": 736.2265625, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.992141330242157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00785865755751729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.83775939941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8060302734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10820, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09941093474626542, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.015169023256748914, + "grad_norm": 0.11252269148826599, + "kept_token_count": 737.228125, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9921895921230316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007810409832745791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1016387939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4094757080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10830, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.10018246918916703, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.015227810945361853, + "grad_norm": 0.09894373267889023, + "kept_token_count": 736.3921875, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9917516589164734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008248344622552395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.2231048583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.12330932617186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10840, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.0995869666337967, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.01726722428575158, + "grad_norm": 0.11559023708105087, + "kept_token_count": 736.9390625, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.992429232597351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007570755667984486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.62498474121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.6226318359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10850, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.09929641783237457, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.015331756882369518, + "grad_norm": 0.11389870196580887, + "kept_token_count": 737.1640625, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9920678198337555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007932182028889656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2417510986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.35906982421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10860, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.10037506371736526, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.01607525646686554, + "grad_norm": 0.09920556098222733, + "kept_token_count": 736.225, + "learning_rate": 2.268679657506519e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920777022838593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007922296784818173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.04527282714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.6708190917969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10870, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10002146363258362, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.01583005813881755, + "grad_norm": 0.11797136813402176, + "kept_token_count": 736.675, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9923452019691468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007654798589646817, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.46067810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.46051330566405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10880, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.10031315013766288, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.015126899909228086, + "grad_norm": 0.10740592330694199, + "kept_token_count": 736.29375, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.992063969373703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007936029322445392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.1720367431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.8768310546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.09979370310902595, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.016145750414580107, + "grad_norm": 0.11158835142850876, + "kept_token_count": 736.5796875, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913224697113037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008677545189857482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.0294250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.66257934570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10900, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.10010634660720825, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.016005503945052624, + "grad_norm": 0.11134129762649536, + "kept_token_count": 736.215625, + "learning_rate": 2.262992765035015e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920285165309906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007971489522606134, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.8060607910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.23004150390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.09972531870007514, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.015547617431730032, + "grad_norm": 0.13162213563919067, + "kept_token_count": 736.89375, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9922887086868286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007711274735629559, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.7333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8103271484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10920, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.09975050613284112, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.016821329947561026, + "grad_norm": 0.14179261028766632, + "kept_token_count": 736.6671875, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9912989795207977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008701029513031245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.87835083007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.5010650634766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10930, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.09984757751226425, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.01543361134827137, + "grad_norm": 0.1448705941438675, + "kept_token_count": 736.90625, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9929856777191162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007014333177357912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0173370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.05446472167966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10940, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.09962602630257607, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.014811109472066163, + "grad_norm": 0.117106594145298, + "kept_token_count": 736.875, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9916124522686005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008387548197060824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.92075500488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4386779785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10950, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.09903022274374962, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.014717836771160364, + "grad_norm": 0.09484205394983292, + "kept_token_count": 737.5046875, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9912736296653748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008726382814347743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.32456665039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.628125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10960, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.09987113326787948, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.01855377359315753, + "grad_norm": 0.1175786554813385, + "kept_token_count": 736.54375, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9922293424606323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007770643662661314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.3227294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.91258544921874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10970, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10015691891312599, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.014873177371919155, + "grad_norm": 0.10440436005592346, + "kept_token_count": 736.3421875, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9921151518821716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007884855382144451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.5487518310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3851379394531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.09982851669192314, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.017117347475141285, + "grad_norm": 0.11093588918447495, + "kept_token_count": 736.625, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9915968418121338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008403149992227554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.51893310546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.45655212402346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 10990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10030243396759034, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.015314703900367021, + "grad_norm": 0.10984405875205994, + "kept_token_count": 736.259375, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9922925293445587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007707462832331658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.11419677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.34229736328126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11000, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09963906705379486, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.01672323290258646, + "grad_norm": 0.10572265088558197, + "kept_token_count": 736.6734375, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9918643116950989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008135668933391571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.05927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7603363037109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11010, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.09970783814787865, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.01604155283421278, + "grad_norm": 0.1122235581278801, + "kept_token_count": 736.8671875, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923299670219421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007670030742883682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8744842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.59324340820314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11020, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.09996879920363426, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.015936762373894454, + "grad_norm": 0.09467049688100815, + "kept_token_count": 736.4765625, + "learning_rate": 2.245843529059e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9914906024932861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008509401138871909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.3806121826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.571630859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11030, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.0997591532766819, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.01436419989913702, + "grad_norm": 0.11693648993968964, + "kept_token_count": 736.65, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9917054533958435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294556941837072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.3147888183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3850891113281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11040, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.09916297793388366, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.01602380145341158, + "grad_norm": 0.10185304284095764, + "kept_token_count": 737.1265625, + "learning_rate": 2.242972554091581e-05, + "loss": 0.02, + "mgd_cos_sim": 0.992011445760727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007988549303263426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8758972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.41175537109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11050, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10033339560031891, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.01643017241731286, + "grad_norm": 0.09483024477958679, + "kept_token_count": 736.43125, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992119950056076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00788004631176591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.04039306640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3664855957031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11060, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.09987388923764229, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.014200747758150101, + "grad_norm": 0.11920157074928284, + "kept_token_count": 736.4625, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9921104490756989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007889555860310793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.67596130371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.1860382080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11070, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10012954548001289, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.016448173299431802, + "grad_norm": 0.14833709597587585, + "kept_token_count": 736.1203125, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9924346446990967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00756534980610013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.87293395996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5972839355469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11080, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.09995132163167, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.014908439945429564, + "grad_norm": 0.11232677102088928, + "kept_token_count": 736.3828125, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9920143723487854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007985618803650141, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.7019073486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4048736572266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11090, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.09945796951651573, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.015215327497571707, + "grad_norm": 0.08960047364234924, + "kept_token_count": 736.6, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9921547770500183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007845241483300925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5022827148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.89910278320315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11100, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.09953379854559899, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.015729228686541318, + "grad_norm": 0.10113628208637238, + "kept_token_count": 736.728125, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9921805441379548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007819459494203328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.1735778808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.20716247558596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11110, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09993084967136383, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.016597061790525912, + "grad_norm": 0.1319032907485962, + "kept_token_count": 736.4609375, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9924131393432617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007586861308664083, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.42539367675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1378234863281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11120, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09842231273651122, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.017886713054031132, + "grad_norm": 0.10278486460447311, + "kept_token_count": 737.4359375, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9912792980670929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008720705658197403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.1364379882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3716217041016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11130, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.09974518939852714, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.015460200607776642, + "grad_norm": 0.09328454732894897, + "kept_token_count": 736.496875, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9921859383583069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007814061269164086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.32248840332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.67369079589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11140, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.10082176774740219, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.013602448534220457, + "grad_norm": 0.09215930104255676, + "kept_token_count": 735.765625, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926301896572113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007369824126362801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9581604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6784606933594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11150, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.1003207191824913, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.017425721045583487, + "grad_norm": 0.1431320607662201, + "kept_token_count": 736.159375, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.992485761642456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007514246553182602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.96653442382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9864990234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11160, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.10004443228244782, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.016959871724247934, + "grad_norm": 0.1121952161192894, + "kept_token_count": 736.0875, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922176957130432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007782316394150257, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.02008666992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.71495666503904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11170, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.10039608851075173, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.01425258107483387, + "grad_norm": 0.12585994601249695, + "kept_token_count": 736.0703125, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9921288609504699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007871160935610532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.46412658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8313507080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11180, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.10045230314135552, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.018824645783752203, + "grad_norm": 0.08947110176086426, + "kept_token_count": 736.0734375, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9917304277420044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008269564621150494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4987487792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7608917236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11190, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.10001679807901383, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.016844256222248076, + "grad_norm": 0.1265319138765335, + "kept_token_count": 736.3578125, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918066918849945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008193294983357191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.52525024414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4707794189453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10041631907224655, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.017274390254169703, + "grad_norm": 0.2279798537492752, + "kept_token_count": 735.9015625, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9919438302516937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00805615559220314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.57977294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.819921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11210, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.10064074397087097, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.014071651175618172, + "grad_norm": 0.1138790175318718, + "kept_token_count": 735.934375, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9924628853797912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007537119369953871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9477172851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.43137817382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11220, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.10013671517372132, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.02026915578171611, + "grad_norm": 0.12846225500106812, + "kept_token_count": 735.9625, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9912308096885681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008769197575747966, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.91357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0187530517578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11230, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.1002787247300148, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.015724755357950926, + "grad_norm": 0.11399997025728226, + "kept_token_count": 735.9921875, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9905348181724548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009465186856687069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.04124145507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3085083007812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11240, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.09969343096017838, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.016681202687323093, + "grad_norm": 0.10635975003242493, + "kept_token_count": 736.3375, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9914986848831177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008501325920224189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.57851257324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1433044433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11250, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.09910697862505913, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.015470649302005767, + "grad_norm": 0.10969263315200806, + "kept_token_count": 737.2734375, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9918991506099701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081008430570364, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.61757507324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5456115722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11260, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.10024670735001565, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.015299252420663833, + "grad_norm": 0.10702474415302277, + "kept_token_count": 736.1515625, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9919020354747772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008097966015338898, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.5639923095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.74867858886716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11270, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.09990263283252716, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.013788045197725297, + "grad_norm": 0.10415248572826385, + "kept_token_count": 736.840625, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9919632732868194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008036734350025655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.20068359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.2218322753906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11280, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.09979110434651375, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.013367849634960294, + "grad_norm": 0.11302553117275238, + "kept_token_count": 736.6734375, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9924513876438141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007548599224537611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.94598388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.9501922607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11290, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09953166022896767, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.01480244267731905, + "grad_norm": 0.09201379120349884, + "kept_token_count": 736.9703125, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9920842111110687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007915791962295771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.4232482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.47706298828126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11300, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.10063463747501374, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.016062442772090436, + "grad_norm": 0.10946029424667358, + "kept_token_count": 736.2296875, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917442560195923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008255742024630308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0893981933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8294311523438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11310, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.09999259561300278, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.014722493756562472, + "grad_norm": 0.12023341655731201, + "kept_token_count": 736.63125, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9920644104480744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007935588527470828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.07143249511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9576110839844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11320, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.09950012117624282, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.01536416532471776, + "grad_norm": 0.0980827584862709, + "kept_token_count": 737.4109375, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924507915973664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007549198344349861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3426025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3793182373047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11330, + "valid_token_count": 818.890625 + }, + { + "actual_token_mask_ratio": 0.09966178014874458, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.017734533362090587, + "grad_norm": 0.10654614120721817, + "kept_token_count": 736.5390625, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9911256849765777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008874315954744815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2244903564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1634124755859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11340, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.09981309920549393, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.01654974976554513, + "grad_norm": 0.11006699502468109, + "kept_token_count": 736.5421875, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9921002268791199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007899781595915556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7323822021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4864837646484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11350, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.09921327829360962, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.016788138449192046, + "grad_norm": 0.13484583795070648, + "kept_token_count": 737.0375, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9910147368907929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008985259663313627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9953643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4734710693359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11360, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.10029733702540397, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.016137309558689594, + "grad_norm": 0.09840698540210724, + "kept_token_count": 736.659375, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.02, + "mgd_cos_sim": 0.992218840122223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007781164161860943, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3971374511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.37969970703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11370, + "valid_token_count": 818.778125 + }, + { + "actual_token_mask_ratio": 0.09958330318331718, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.015999641735106707, + "grad_norm": 0.12064464390277863, + "kept_token_count": 736.9359375, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918572008609772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008142795413732529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.6794860839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.002392578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11380, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.09987200349569321, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.016998468339443205, + "grad_norm": 0.10958129167556763, + "kept_token_count": 736.715625, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9918578684329986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00814214712008834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.9792999267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.5606323242188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.10026721432805061, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.016811282467097045, + "grad_norm": 0.11176101863384247, + "kept_token_count": 736.3171875, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9924304127693176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007569583505392075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6962524414063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1079803466797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11400, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.09974638223648072, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.01658496670424938, + "grad_norm": 0.12236453592777252, + "kept_token_count": 736.95, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9916692912578583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008330695610493422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9700988769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.87615661621095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11410, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.1002173200249672, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.014093613438308238, + "grad_norm": 0.12846101820468903, + "kept_token_count": 736.2171875, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9924677968025207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007532210927456618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.04954223632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1286682128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11420, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09960473105311393, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.014069775585085154, + "grad_norm": 0.09141427278518677, + "kept_token_count": 736.484375, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927612900733948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072387074120342735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6338409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11430, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.09982099533081054, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.013405665475875139, + "grad_norm": 0.09131818264722824, + "kept_token_count": 736.809375, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9924899995326996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007510009966790676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0665710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.8886962890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11440, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.10004981607198715, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.01548419687896967, + "grad_norm": 0.0892668217420578, + "kept_token_count": 736.2109375, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9924334168434144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007566580455750227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0486297607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5114990234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11450, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.10028555989265442, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.016030693612992763, + "grad_norm": 0.13640235364437103, + "kept_token_count": 736.2296875, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9916940808296204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008305921033024789, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.76023254394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7854309082031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.09971729144454003, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.014084869250655174, + "grad_norm": 0.1218036338686943, + "kept_token_count": 736.828125, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9921863317489624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00781367514282465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9424163818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0913482666016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11470, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.09949617758393288, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.015030030719935894, + "grad_norm": 0.1131347194314003, + "kept_token_count": 736.5109375, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922162175178528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007783791515976191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6070922851562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7161560058594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11480, + "valid_token_count": 817.8890625 + }, + { + "actual_token_mask_ratio": 0.09954805746674537, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.015671659633517264, + "grad_norm": 0.11913295835256577, + "kept_token_count": 736.915625, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9924761891365051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007523817289620638, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.79527587890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0782135009766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11490, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.10031797289848328, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.015164382709190249, + "grad_norm": 0.12318272888660431, + "kept_token_count": 736.1390625, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922796487808228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00772036612033844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.23012390136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0195617675781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11500, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09950520247220992, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.016249438375234605, + "grad_norm": 0.13923977315425873, + "kept_token_count": 736.8484375, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9921398222446441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007860175333917141, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.38340759277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.2590637207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11510, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10059159025549888, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.0178211129270494, + "grad_norm": 0.09521190077066422, + "kept_token_count": 736.1015625, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9919114589691163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008088550250977277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6172790527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.32347412109374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11520, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.09980457574129105, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.018307488597929477, + "grad_norm": 0.10388430953025818, + "kept_token_count": 736.6140625, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9925361216068268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007463888637721539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.5084228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5450897216797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11530, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.10033479034900665, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.017657206300646065, + "grad_norm": 0.16626885533332825, + "kept_token_count": 736.2953125, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9921956300735474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007804380822926759, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9960479736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9623779296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11540, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.10002893060445786, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.017736068833619355, + "grad_norm": 0.10422540456056595, + "kept_token_count": 736.4640625, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9921712160110474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007828790321946144, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.67881774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.55809020996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11550, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.10034837946295738, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.016562940552830696, + "grad_norm": 0.1335975080728531, + "kept_token_count": 736.4703125, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9925009608268738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007499051466584206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5408660888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0096099853516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11560, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.09968468248844146, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.015192071348428727, + "grad_norm": 0.12864789366722107, + "kept_token_count": 736.740625, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922494888305664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007750502880662679, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.5435455322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.41383666992186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11570, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.1007104493677616, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.017312853410840035, + "grad_norm": 0.09035255014896393, + "kept_token_count": 735.765625, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9924988806247711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007501116301864385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.8602783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.93101196289064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11580, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.10030036270618439, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.017588441353291275, + "grad_norm": 0.09541179984807968, + "kept_token_count": 736.003125, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9927969813346863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072030358947813514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2996551513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.37073974609376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11590, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.10048827975988388, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.015758526604622604, + "grad_norm": 0.12121853977441788, + "kept_token_count": 736.234375, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9917569518089294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008243053127080202, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4046173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.91368408203124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11600, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.0997553214430809, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.018271594773977995, + "grad_norm": 0.1275264024734497, + "kept_token_count": 736.971875, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9914402782917022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00855971546843648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2684600830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.29156188964845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11610, + "valid_token_count": 818.63125 + }, + { + "actual_token_mask_ratio": 0.10015566945075989, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.019345520436763762, + "grad_norm": 0.11439778655767441, + "kept_token_count": 736.3796875, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9912722826004028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008727713860571385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4497344970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4158630371094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11620, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.10027719587087632, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.019430086016654968, + "grad_norm": 0.16645878553390503, + "kept_token_count": 736.2421875, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9919852316379547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008014774695038796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.0166900634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.55262451171876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.0993529014289379, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.015762438904494046, + "grad_norm": 0.1319771260023117, + "kept_token_count": 736.85, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9924086809158326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007591315917670727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.85565795898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1773315429688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11640, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.09971695616841317, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.016138625890016557, + "grad_norm": 0.11152330040931702, + "kept_token_count": 736.859375, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9925588250160218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441171538084745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.32203369140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1483978271484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11650, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.10010807290673256, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.014842118136584759, + "grad_norm": 0.10409950464963913, + "kept_token_count": 736.3953125, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927698493003845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007230153866112232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.80848083496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2657073974609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11660, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.0998982772231102, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.01591474823653698, + "grad_norm": 0.13338603079319, + "kept_token_count": 736.465625, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9924095928668976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007590392325073481, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.78590087890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.67820129394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11670, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.10032685697078705, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.014458747487515211, + "grad_norm": 0.11223176121711731, + "kept_token_count": 736.153125, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9920325934886932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007967409025877715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5795654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7338623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11680, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10014900267124176, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.015146713517606259, + "grad_norm": 0.13702934980392456, + "kept_token_count": 736.3859375, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9924048066139222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075951824896037575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.68528442382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5033203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09983205497264862, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.017900626454502345, + "grad_norm": 0.1279439777135849, + "kept_token_count": 736.5171875, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9923559427261353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007644064072519541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.38421325683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.7622406005859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11700, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.10072088316082954, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.015559318475425243, + "grad_norm": 0.11541658639907837, + "kept_token_count": 735.590625, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.992477148771286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075228272937238215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.4279022216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.17640075683596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11710, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.09991964921355248, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.0146244952455163, + "grad_norm": 0.11107635498046875, + "kept_token_count": 736.4078125, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9923164486885071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007683547865599394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1129547119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.3741455078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11720, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.09970536455512047, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.016975655406713485, + "grad_norm": 0.15210793912410736, + "kept_token_count": 736.828125, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9920169293880463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007983067724853753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.9411590576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9689208984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11730, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.09968365803360939, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.016157980635762216, + "grad_norm": 0.09139961749315262, + "kept_token_count": 736.778125, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922902226448059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007709774561226368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.73375244140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.12070617675784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11740, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09984425082802773, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.017952132131904365, + "grad_norm": 0.11426601558923721, + "kept_token_count": 736.5125, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9923358678817749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076641284860670565, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.16490783691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.11961364746094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11750, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.09945058599114417, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.016717169806361197, + "grad_norm": 0.11789786070585251, + "kept_token_count": 737.025, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9920991003513336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007900897227227689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.1938079833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.895166015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11760, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.10048876851797103, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.015531361103057861, + "grad_norm": 0.11885207891464233, + "kept_token_count": 736.0, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921005427837372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007899463921785355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7862213134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0639190673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11770, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09976859018206596, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.013157903589308262, + "grad_norm": 0.10756643116474152, + "kept_token_count": 736.8484375, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9925633192062377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007436684332787991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.36708679199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.57447509765626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.10003790035843849, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.01585213476791978, + "grad_norm": 0.11562713235616684, + "kept_token_count": 736.41875, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9921221733093262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007877833023667336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.47772521972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.93052978515624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11790, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.10011218786239624, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.01708314549177885, + "grad_norm": 0.1447124183177948, + "kept_token_count": 736.50625, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919860124588012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008013982605189086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.5490447998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.22242126464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11800, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.1004375584423542, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.01659141629934311, + "grad_norm": 0.1269363909959793, + "kept_token_count": 736.115625, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9924472749233246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075527200475335125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7883728027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.42909545898436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11810, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.10015809312462806, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.016581878066062927, + "grad_norm": 0.13050834834575653, + "kept_token_count": 736.2140625, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9922135412693024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077864454127848145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3306182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7635772705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11820, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.09990731701254844, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.016754483617842198, + "grad_norm": 0.11074337363243103, + "kept_token_count": 736.6484375, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9918225586414338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081774334423244, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.2843933105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68101196289064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11830, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.10037119761109352, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.017578718438744544, + "grad_norm": 0.13110311329364777, + "kept_token_count": 735.928125, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9918795526027679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008120457362383603, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3921173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1368865966797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11840, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.10039505809545517, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.01616417970508337, + "grad_norm": 0.13243766129016876, + "kept_token_count": 736.1296875, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9918648481369019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008135140780359507, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4646850585938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.65479431152346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11850, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.0998598113656044, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.014791107457131147, + "grad_norm": 0.10377760231494904, + "kept_token_count": 736.6171875, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9921055555343627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007894445303827524, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.34205322265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3143798828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11860, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09934192523360252, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.016472384333610535, + "grad_norm": 0.10396505147218704, + "kept_token_count": 737.2625, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9925280094146729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074719835072755815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7074920654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.83839721679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11870, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.09954836666584015, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.016526336036622524, + "grad_norm": 0.15425364673137665, + "kept_token_count": 736.7796875, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923233926296234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007676608860492706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4671203613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.03846740722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11880, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.09963094741106034, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.016185926087200642, + "grad_norm": 0.13881036639213562, + "kept_token_count": 736.85625, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920580327510834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007941954769194126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1055908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9089050292969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11890, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10034801065921783, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.015727965999394654, + "grad_norm": 0.13155481219291687, + "kept_token_count": 736.284375, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9919622004032135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008037794101983308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.7785278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.23585815429686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11900, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.10021955296397209, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.016604919359087943, + "grad_norm": 0.12351013720035553, + "kept_token_count": 736.2234375, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9915663957595825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008433606661856174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.366650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6601226806641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11910, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.09973217472434044, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.01617012470960617, + "grad_norm": 0.11404268443584442, + "kept_token_count": 736.73125, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9919493138790131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008050696272403002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.82423095703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.09925842285156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11920, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.09928228557109833, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.019088547490537167, + "grad_norm": 0.1276242733001709, + "kept_token_count": 736.7578125, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9924332320690155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00756676122546196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5227935791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.49354858398436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11930, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.09991947039961815, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.01894457349553704, + "grad_norm": 0.12672902643680573, + "kept_token_count": 736.5140625, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9921537637710571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00784623296931386, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1127410888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.77304992675784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11940, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.10051645338535309, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.01577363219112158, + "grad_norm": 0.09276913851499557, + "kept_token_count": 736.1328125, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9922819316387177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077180730178952215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.3442443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5946075439453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.10025810152292251, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.018029558565467597, + "grad_norm": 0.12428172677755356, + "kept_token_count": 736.109375, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9922666549682617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007733334135264158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.2173583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0649353027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11960, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.10004687458276748, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.015375636145472526, + "grad_norm": 0.10592176765203476, + "kept_token_count": 736.5671875, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923731684684753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007626860309392214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5516815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.91566772460936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11970, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.10075004547834396, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.01613298710435629, + "grad_norm": 0.15303772687911987, + "kept_token_count": 736.1171875, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922873795032501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007712609507143498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.68458251953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.21629638671874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11980, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.10017757713794709, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.014145651087164878, + "grad_norm": 0.12417830526828766, + "kept_token_count": 736.1359375, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9923726439476013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007627358753234148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9403503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.1296417236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 11990, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09999047666788101, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.01629795813933015, + "grad_norm": 0.09375204890966415, + "kept_token_count": 736.578125, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9925250828266143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00747491791844368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.82501525878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.41639709472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12000, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10104203298687935, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.018314696941524745, + "grad_norm": 0.13214589655399323, + "kept_token_count": 735.4890625, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9920632421970368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007936750259250402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.85145263671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6660614013672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12010, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10050636231899261, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.013832790916785598, + "grad_norm": 0.13229553401470184, + "kept_token_count": 736.025, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9923273205757142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007672681380063296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.9427429199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9371765136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12020, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.09981630444526672, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.015041150897741318, + "grad_norm": 0.10892248153686523, + "kept_token_count": 736.26875, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9918948650360108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810513123869896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.4228973388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4507690429688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12030, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.0998822145164013, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.015730393305420875, + "grad_norm": 0.10210593044757843, + "kept_token_count": 736.4453125, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9920112192630768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0079887880012393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.1691009521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7924285888672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.10009258165955544, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.01649971790611744, + "grad_norm": 0.09291885793209076, + "kept_token_count": 736.484375, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922411024570466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007758889347314835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.30177612304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4678161621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.0995500311255455, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.01781811947003007, + "grad_norm": 0.12048515677452087, + "kept_token_count": 736.9734375, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9919468104839325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008053196780383586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6361358642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3551300048828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12060, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.09994639158248901, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.013947471603751182, + "grad_norm": 0.09726201742887497, + "kept_token_count": 736.31875, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9920891404151917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007910865545272826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5202209472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9006652832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12070, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.0996064193546772, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.01504724146798253, + "grad_norm": 0.09214981645345688, + "kept_token_count": 736.959375, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.991941112279892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008058892842382193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.3578277587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2977722167969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12080, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.10024864599108696, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.01402897210791707, + "grad_norm": 0.0958392322063446, + "kept_token_count": 736.296875, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9920219898223877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007978019956499337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.70245056152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.84043273925784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12090, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09996778890490532, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.01647114474326372, + "grad_norm": 0.08838655799627304, + "kept_token_count": 736.4890625, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922386288642884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007761386688798666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4191833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.07432250976564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12100, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1005482703447342, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.0161401997320354, + "grad_norm": 0.0998648852109909, + "kept_token_count": 735.69375, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992086660861969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007913353387266397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5580230712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.94777221679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12110, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.09933944195508956, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.015563902538269758, + "grad_norm": 0.11832541227340698, + "kept_token_count": 737.1890625, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921585083007812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007841495331376791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.186279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.5274932861328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12120, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.0998257890343666, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.01624338086694479, + "grad_norm": 0.14662238955497742, + "kept_token_count": 736.7328125, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9916722416877747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00832776091992855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.3957244873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6564636230469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12130, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.09980686902999877, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.015487146750092506, + "grad_norm": 0.12053321301937103, + "kept_token_count": 736.6515625, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9918268144130706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008173182886093856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.67213745117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1446960449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12140, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.09934230074286461, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.014859973266720772, + "grad_norm": 0.09178175032138824, + "kept_token_count": 736.99375, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927072346210479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007292760442942381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.8897705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.69195251464845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12150, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.09975536316633224, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.016305055376142262, + "grad_norm": 0.1208617240190506, + "kept_token_count": 736.5125, + "learning_rate": 2.078375988416095e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9925187230110168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007481264043599367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0925231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.11359252929685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12160, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.10008381977677346, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.014964440278708935, + "grad_norm": 0.11316359788179398, + "kept_token_count": 736.146875, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9920985460281372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007901455648243428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.6473846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.29375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12170, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.09955460503697396, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.01753772273659706, + "grad_norm": 0.11295212060213089, + "kept_token_count": 736.8828125, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9925717413425446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074282740242779255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5188873291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5500244140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12180, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10012129992246628, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.014505134150385856, + "grad_norm": 0.16727757453918457, + "kept_token_count": 736.440625, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.991738247871399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261754456907511, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.1818145751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5333221435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12190, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.1000776894390583, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.016730606276541948, + "grad_norm": 0.10949410498142242, + "kept_token_count": 736.3046875, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9925633907318115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007436604890972376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.5567596435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6911682128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12200, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.0999783918261528, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.018426398932933807, + "grad_norm": 0.15237051248550415, + "kept_token_count": 736.39375, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9922304630279541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007769542559981346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.55971069335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3298400878906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12210, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.0997846782207489, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.014511491730809211, + "grad_norm": 0.13653843104839325, + "kept_token_count": 736.634375, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9926265001296997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007373505737632513, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.98891296386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4131378173828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12220, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.10008663386106491, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.01605608742684126, + "grad_norm": 0.08934566378593445, + "kept_token_count": 736.3390625, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.992482590675354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075174118392169476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.6179443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2974487304688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12230, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09998009577393532, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.01655208533629775, + "grad_norm": 0.1366182565689087, + "kept_token_count": 736.3359375, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919837057590485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008016299456357956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4000701904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.93287353515626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12240, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.09974120035767556, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.014195853471755981, + "grad_norm": 0.09918365627527237, + "kept_token_count": 736.7, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9928105533123016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007189433090388775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.1864959716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.45653076171874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.09981408640742302, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.016283561103045942, + "grad_norm": 0.11417534947395325, + "kept_token_count": 736.6171875, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9909622550010682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009037757199257613, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.5046783447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.93108825683595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12260, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.099834343791008, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.01783473351970315, + "grad_norm": 0.10120552778244019, + "kept_token_count": 736.8875, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9920935690402984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007906424533575773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.07909240722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9507781982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12270, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.09976821094751358, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.013781383726745844, + "grad_norm": 0.11938432604074478, + "kept_token_count": 736.9921875, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9921835422515869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00781646417453885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6739166259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.58363037109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12280, + "valid_token_count": 818.6703125 + }, + { + "actual_token_mask_ratio": 0.10026830732822418, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013337615318596363, + "grad_norm": 0.1011258140206337, + "kept_token_count": 736.4796875, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9924306809902191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007569303456693888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.9017852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7502105712891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12290, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.10006342381238938, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.018064860440790653, + "grad_norm": 0.13609206676483154, + "kept_token_count": 736.634375, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9925376772880554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007462313864380121, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8925415039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4334259033203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12300, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.09979111850261688, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.016606989409774543, + "grad_norm": 0.1187213659286499, + "kept_token_count": 736.5984375, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9921354830265046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007864530850201844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1167785644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5214599609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12310, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.0997076041996479, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.015702633094042538, + "grad_norm": 0.10870348662137985, + "kept_token_count": 736.425, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.992122209072113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007877797912806272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.91394958496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.53818969726564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12320, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.09974436014890671, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.014471337758004665, + "grad_norm": 0.16393746435642242, + "kept_token_count": 736.978125, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9920704662799835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007929528318345546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.78098449707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.66787109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12330, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.10014771893620492, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.015560649614781142, + "grad_norm": 0.10766813904047012, + "kept_token_count": 736.1890625, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9919932842254638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0080067191272974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.0943267822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2041351318359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12340, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.09925071150064468, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.017024369537830354, + "grad_norm": 0.09661342948675156, + "kept_token_count": 737.0, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922945439815521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077054501511156556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9323425292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4501129150391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10074787512421608, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.01594453016296029, + "grad_norm": 0.10263142734766006, + "kept_token_count": 736.190625, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9921512842178345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007848703395575285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2623687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0482940673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12360, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.0988021269440651, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.014524392317980528, + "grad_norm": 0.09547140449285507, + "kept_token_count": 737.50625, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9917052209377288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00829478893429041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.0641296386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 391.82965087890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12370, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.09971413537859916, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.01401991923339665, + "grad_norm": 0.08225234597921371, + "kept_token_count": 736.94375, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9923239111900329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00767609691247344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2320892333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.1411651611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12380, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1005267433822155, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.015032089967280626, + "grad_norm": 0.10126476734876633, + "kept_token_count": 736.0171875, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924421370029449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075578693300485614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.24542541503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.5544006347656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12390, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.10082854554057122, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.014969277009367944, + "grad_norm": 0.14713943004608154, + "kept_token_count": 736.1453125, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9923932254314423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007606792263686657, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.53187255859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.25486145019534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12400, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.09980951920151711, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.013403852842748165, + "grad_norm": 0.10729682445526123, + "kept_token_count": 737.0015625, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9921923220157624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007807677239179611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.18276062011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.80408630371096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12410, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.0998228743672371, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.014349793642759323, + "grad_norm": 0.1077028214931488, + "kept_token_count": 736.4609375, + "learning_rate": 2.038477994913242e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927939176559448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072060924023389815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.98964233398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.54896545410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12420, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.10034580081701279, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.018155866488814353, + "grad_norm": 0.10537251085042953, + "kept_token_count": 735.990625, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.992134177684784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00786582762375474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.8138153076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.64751892089845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12430, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.10028324499726296, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.013499338226392866, + "grad_norm": 0.12116068601608276, + "kept_token_count": 736.5390625, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9927041411399842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007295853458344936, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.48614501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.1802459716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12440, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.0994940847158432, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.014289718493819237, + "grad_norm": 0.12143240123987198, + "kept_token_count": 737.01875, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9921956539154053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00780434524640441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9511413574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9899627685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12450, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.1005832776427269, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.013073874916881324, + "grad_norm": 0.08126217126846313, + "kept_token_count": 736.1390625, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9927043080329895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007295699045062065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.6000518798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2154174804688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12460, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.09957238361239433, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.016947594843804835, + "grad_norm": 0.12232565134763718, + "kept_token_count": 736.934375, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9923413097858429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007658698502928019, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.50311889648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.20701293945314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12470, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.10018517374992371, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.01337112272158265, + "grad_norm": 0.10503585636615753, + "kept_token_count": 736.103125, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9922710239887238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007728977501392365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.5283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2562591552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12480, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.10005707815289497, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016461155842989683, + "grad_norm": 0.09015554934740067, + "kept_token_count": 736.440625, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9925962626934052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074037425220012665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.5226776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3006195068359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12490, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10062030777335167, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.014696329273283482, + "grad_norm": 0.12632213532924652, + "kept_token_count": 735.853125, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922316551208497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00776835922151804, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9830352783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2987854003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12500, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.09965822249650955, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.016646123863756655, + "grad_norm": 0.13020315766334534, + "kept_token_count": 736.984375, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.992010498046875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007989500742405653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8930145263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.8892547607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12510, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.0998485378921032, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.013079185504466295, + "grad_norm": 0.10088393092155457, + "kept_token_count": 736.4921875, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9923807799816131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076192268170416355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8689788818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.37205505371094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12520, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.09997258260846138, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.016019020229578018, + "grad_norm": 0.1428939700126648, + "kept_token_count": 736.603125, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9924286782741547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007571326289325953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9267517089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2615997314453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12530, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.09953673109412194, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.014085338264703751, + "grad_norm": 0.11851093918085098, + "kept_token_count": 736.8203125, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.018, + "mgd_cos_sim": 0.992088234424591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791176874190569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4426025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6786193847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.10012378618121147, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.015537601243704558, + "grad_norm": 0.13178858160972595, + "kept_token_count": 736.1609375, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920859813690186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791401881724596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.3789398193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.17661743164064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12550, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.10028859525918961, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.017700227443128824, + "grad_norm": 0.1033441424369812, + "kept_token_count": 736.0015625, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9926587998867035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007341202907264232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7770050048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5063903808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12560, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.1003495566546917, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.015388658735901117, + "grad_norm": 0.13388584554195404, + "kept_token_count": 736.1578125, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9921945333480835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007805459666997194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6839660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4744903564453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12570, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.09963702484965324, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.017240460123866798, + "grad_norm": 0.1332400143146515, + "kept_token_count": 736.815625, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9928812503814697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071187449619174005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.81459655761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.62547302246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12580, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10020973160862923, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.015017411392182111, + "grad_norm": 0.10940999537706375, + "kept_token_count": 736.33125, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9925332427024841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074667539447546005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.14794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1473022460938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09981918260455132, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.014107507187873125, + "grad_norm": 0.11095760762691498, + "kept_token_count": 736.6640625, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9920972406864166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00790276750922203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6868560791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 392.7431915283203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12600, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.09979502335190774, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.012793876044452191, + "grad_norm": 0.12259601801633835, + "kept_token_count": 736.528125, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9925597965717315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007440208736807108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.16469116210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9032958984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12610, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.10025589838624001, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.016111503168940545, + "grad_norm": 0.11447244882583618, + "kept_token_count": 736.1828125, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9919239401817321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008076062891632318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0364715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.97935791015624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12620, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.09980947449803353, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.01626710770651698, + "grad_norm": 0.10623767971992493, + "kept_token_count": 736.403125, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.992755651473999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007244346477091312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.5891571044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.37665405273435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.09973526373505592, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.016913378424942494, + "grad_norm": 0.12150008231401443, + "kept_token_count": 736.4828125, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9927217125892639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007278286758810282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.84954528808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7280609130859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12640, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.10045598447322845, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.014424816705286503, + "grad_norm": 0.10156263411045074, + "kept_token_count": 735.9078125, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927497684955597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007250220701098442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.98345947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.33436584472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12650, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.099597530066967, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.015256512723863126, + "grad_norm": 0.10490144789218903, + "kept_token_count": 736.7671875, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9929828524589539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007017149776220322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.95879516601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.45641174316404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.10021139234304428, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.015879850275814532, + "grad_norm": 0.10459102690219879, + "kept_token_count": 736.1359375, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9923875093460083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007612492889165878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0364074707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4562164306641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12670, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10023742988705635, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.014688164554536342, + "grad_norm": 0.09484335035085678, + "kept_token_count": 736.5953125, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9925945162773132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074054933153092865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.51512756347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.03745727539064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12680, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.10104619860649108, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.016580027434974907, + "grad_norm": 0.10759266465902328, + "kept_token_count": 735.9125, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9925810396671295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007418967690318823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5973846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.00791015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12690, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.10058998167514802, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.01472906144335866, + "grad_norm": 0.10767021775245667, + "kept_token_count": 735.8578125, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9927692532539367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007230746746063233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.2810424804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5259674072266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12700, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10003561303019523, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.015275860391557216, + "grad_norm": 0.10140024870634079, + "kept_token_count": 736.5328125, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.992804902791977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007195109501481056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.69818115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7166473388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12710, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.10092402994632721, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.017166062816977502, + "grad_norm": 0.10126049071550369, + "kept_token_count": 735.5046875, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9926300585269928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007369942311197519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.24232482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7138305664063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12720, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.09961957037448883, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.01745218215510249, + "grad_norm": 0.11441338062286377, + "kept_token_count": 736.6125, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9924852550029755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007514756731688976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.992333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7892333984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12730, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.0999268837273121, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.014226861577481032, + "grad_norm": 0.09993354976177216, + "kept_token_count": 736.51875, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9929613769054413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038605120033026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.43402404785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.6640289306641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12740, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09928186386823654, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.01531759025529027, + "grad_norm": 0.09187890589237213, + "kept_token_count": 736.859375, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9929995954036712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007000405620783567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0827911376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.35789184570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12750, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.10060499086976052, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.016613937588408588, + "grad_norm": 0.09410610795021057, + "kept_token_count": 736.0640625, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992355477809906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00764452600851655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.73359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4370910644531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12760, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09989919513463974, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.017317530792206527, + "grad_norm": 0.10746299475431442, + "kept_token_count": 736.4921875, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923375964164733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007662412896752358, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.03965454101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0085205078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12770, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09916447773575783, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.015474947448819876, + "grad_norm": 0.12287575006484985, + "kept_token_count": 737.065625, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9924295246601105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007570480555295944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6850891113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.69005432128904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12780, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09920028522610665, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.016824133228510618, + "grad_norm": 0.10671546310186386, + "kept_token_count": 736.9296875, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917229413986206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008277066983282566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1337921142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8895568847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12790, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.10058558285236359, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.015586904529482127, + "grad_norm": 0.10941903293132782, + "kept_token_count": 735.6609375, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927124619483948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007287554442882538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8170135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9701721191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12800, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.10043835192918778, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.0140145193785429, + "grad_norm": 0.09477914869785309, + "kept_token_count": 736.128125, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.992503535747528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007496452610939741, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.05516357421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9152374267578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12810, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09952854439616203, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.015857228077948093, + "grad_norm": 0.09600348770618439, + "kept_token_count": 737.071875, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9925845861434937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007415424380451441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9072326660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1920562744141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12820, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.09967887327075005, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.015406341478228568, + "grad_norm": 0.10858570039272308, + "kept_token_count": 737.059375, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.992108815908432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007891196757555008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.94756469726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.78905029296874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12830, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.09982116371393204, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015332949813455343, + "grad_norm": 0.10969565063714981, + "kept_token_count": 736.7296875, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9917770802974701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008222927618771792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.25391845703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8360229492188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12840, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.0998292438685894, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.017576827108860014, + "grad_norm": 0.13015589118003845, + "kept_token_count": 736.7515625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9916561603546142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008343837317079305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0023559570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.37216796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12850, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.10014911219477654, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.014817552641034126, + "grad_norm": 0.1023518368601799, + "kept_token_count": 736.40625, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9925183773040771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007481608912348748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.35076599121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.77491149902346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12860, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.09914127141237258, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.01585700800642371, + "grad_norm": 0.12173878401517868, + "kept_token_count": 737.0234375, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9924186944961548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007581292744725942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4941833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1766357421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12870, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.0996039479970932, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.014770220499485731, + "grad_norm": 0.10428772866725922, + "kept_token_count": 736.5671875, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9922284662723542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007771525532007217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.22003784179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.18580322265626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12880, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.10003540441393852, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.01650725556537509, + "grad_norm": 0.14457669854164124, + "kept_token_count": 736.2640625, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9926550984382629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007344894763082266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8675079345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.26438903808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12890, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.10019669830799102, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.013331558834761382, + "grad_norm": 0.08633100986480713, + "kept_token_count": 736.2609375, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.99253351688385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007466485071927309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.7560974121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.84368896484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.10046954452991486, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.014513766579329967, + "grad_norm": 0.10584378987550735, + "kept_token_count": 736.075, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9923334002494812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007666590716689825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.24546203613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4802307128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.10067526325583458, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.015770839620381595, + "grad_norm": 0.11605995148420334, + "kept_token_count": 735.8546875, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9928137958049774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071862020529806616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.32505187988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8744232177734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12920, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09910294786095619, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.017342964839190244, + "grad_norm": 0.11691901087760925, + "kept_token_count": 737.290625, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9923848390579224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076151610352098945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.9195495605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7606597900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12930, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10016469284892082, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.01485923184081912, + "grad_norm": 0.11924679577350616, + "kept_token_count": 736.440625, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9924203872680664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007579630427062512, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.2439239501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.5432952880859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12940, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.10010909959673882, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.016215211898088455, + "grad_norm": 0.10755689442157745, + "kept_token_count": 736.6203125, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920582473278046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007941754814237357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.32115783691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.74713134765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12950, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.10043325424194335, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.015602819714695215, + "grad_norm": 0.09514245390892029, + "kept_token_count": 736.04375, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922861993312836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007713790982961655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.6885009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.94675903320314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12960, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09937005564570427, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.013993757218122483, + "grad_norm": 0.10142537206411362, + "kept_token_count": 736.7015625, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9925351679325104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007464840728789568, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.36199340820315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7231506347656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12970, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.10021617934107781, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.014786860533058643, + "grad_norm": 0.109957255423069, + "kept_token_count": 736.3796875, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9924104750156403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007589524146169424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2764831542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.40675964355466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12980, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.0990861214697361, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.01483048927038908, + "grad_norm": 0.13671858608722687, + "kept_token_count": 737.2328125, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9922130227088928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007786969561129809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.5032531738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.39720764160154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 12990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.09961670264601707, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.01736123003065586, + "grad_norm": 0.11612936109304428, + "kept_token_count": 736.6875, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9922219038009643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007778105698525906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3793243408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.57391967773435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13000, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10046207457780838, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.014998095482587815, + "grad_norm": 0.11837746202945709, + "kept_token_count": 736.1625, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9921589672565461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007841045781970025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9473876953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.27820739746096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13010, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.09954954087734222, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.01540251486003399, + "grad_norm": 0.10348325967788696, + "kept_token_count": 736.9265625, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.992254227399826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007745775114744902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6723937988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9220306396484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13020, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.09988437592983246, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.014848832599818706, + "grad_norm": 0.13006286323070526, + "kept_token_count": 736.53125, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927762150764465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007223785854876041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.57391967773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.23984985351564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13030, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09972068592905999, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.014189671911299228, + "grad_norm": 0.0776362344622612, + "kept_token_count": 736.5546875, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992512309551239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007487688213586807, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.56422119140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4975280761719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13040, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.09960168823599816, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.016499888245016338, + "grad_norm": 0.14415375888347626, + "kept_token_count": 736.865625, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9926801145076751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073198894970119, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7913299560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.63624267578126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13050, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.10002574175596238, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.01426067678257823, + "grad_norm": 0.12212236225605011, + "kept_token_count": 736.3796875, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9929695665836334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007030446082353592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0869110107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3711730957031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13060, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09943350926041603, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.01299413563683629, + "grad_norm": 0.09845487028360367, + "kept_token_count": 737.1484375, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.992549604177475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007450398895889521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.85252075195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1990692138672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13070, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.10001818239688873, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.016675790958106517, + "grad_norm": 0.13002391159534454, + "kept_token_count": 736.4890625, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9923360407352447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007663968484848738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0062744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5794952392578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13080, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.09922639280557632, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.016778345685452223, + "grad_norm": 0.12688149511814117, + "kept_token_count": 737.528125, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9930291414260864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006970875803381205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9603607177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5461090087891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13090, + "valid_token_count": 818.7703125 + }, + { + "actual_token_mask_ratio": 0.10014025568962097, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.014142661960795522, + "grad_norm": 0.13394968211650848, + "kept_token_count": 736.265625, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9925687432289123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007431267853826284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.77846069335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7842498779297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13100, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.10020191445946694, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.01686993455514312, + "grad_norm": 0.12240204215049744, + "kept_token_count": 736.1390625, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9923397123813629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007660288922488689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.3474639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1901428222656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13110, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.10026160329580307, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.01389104463160038, + "grad_norm": 0.1001732349395752, + "kept_token_count": 736.3515625, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9928334712982178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071665307506918905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.18511657714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6774932861328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13120, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.09969133362174035, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.015831335447728634, + "grad_norm": 0.1220887079834938, + "kept_token_count": 736.8234375, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9919930696487427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008006932493299247, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.13551635742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.84312438964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13130, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10068074688315391, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.013975793682038783, + "grad_norm": 0.11649945378303528, + "kept_token_count": 735.95625, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927070319652558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00729296775534749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.7542724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4683135986328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.10017623826861381, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.015044211875647306, + "grad_norm": 0.11118648201227188, + "kept_token_count": 736.2578125, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9923831343650817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076168617233633995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.93211975097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8636810302734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13150, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.10041027069091797, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.013800016324967145, + "grad_norm": 0.0979514867067337, + "kept_token_count": 736.2703125, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9924003839492798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075996160507202145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.2673614501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0463104248047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13160, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.09963377490639687, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.016596125904470682, + "grad_norm": 0.09935218840837479, + "kept_token_count": 736.8765625, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9924474775791168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007552529033273458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0458526611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0584381103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13170, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.09962408840656281, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.014565172418951989, + "grad_norm": 0.09991975873708725, + "kept_token_count": 736.5859375, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9923861920833588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007613800745457411, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2440155029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.78883056640626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13180, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.10028488114476204, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.017064567562192677, + "grad_norm": 0.11009150743484497, + "kept_token_count": 736.0890625, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919733822345733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008026611618697643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.9565093994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5719299316406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13190, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.09999416917562484, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.013685251213610172, + "grad_norm": 0.10471885651350021, + "kept_token_count": 736.2046875, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9922260999679565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007773897890001535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3025787353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8328399658203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13200, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.10011829584836959, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.01560930022969842, + "grad_norm": 0.09604468196630478, + "kept_token_count": 736.31875, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9926149308681488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007385062705725432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4147521972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5656280517578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13210, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09944865480065346, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.01458605332300067, + "grad_norm": 0.10466337949037552, + "kept_token_count": 736.5125, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9921447098255157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007855277974158525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.73081665039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7899566650391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13220, + "valid_token_count": 817.846875 + }, + { + "actual_token_mask_ratio": 0.09996780529618263, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.015654630772769452, + "grad_norm": 0.10246129333972931, + "kept_token_count": 736.5328125, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9928428351879119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007157155871391296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9311889648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.18427124023435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13230, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.10000107958912849, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.013584488816559315, + "grad_norm": 0.09345262497663498, + "kept_token_count": 736.25, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.992505532503128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007494466938078403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.668017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.899560546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13240, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.09989123046398163, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.015306833386421203, + "grad_norm": 0.0999554842710495, + "kept_token_count": 736.559375, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923074543476105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00769254108890891, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.667626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9406402587891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13250, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.0997015506029129, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.01744592310860753, + "grad_norm": 0.10993828624486923, + "kept_token_count": 737.0078125, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9926626980304718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007337287534028292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0620056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9012176513672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13260, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.10051722452044487, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.014629646996036172, + "grad_norm": 0.09304075688123703, + "kept_token_count": 735.4375, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927741944789886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007225804589688778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.5292114257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3168029785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13270, + "valid_token_count": 817.621875 + }, + { + "actual_token_mask_ratio": 0.09982550740242005, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.014922415278851986, + "grad_norm": 0.10135887563228607, + "kept_token_count": 736.809375, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9925642311573029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007435760833323002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.41722412109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.76402893066404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13280, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.10031994134187698, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.01609726371243596, + "grad_norm": 0.12332190573215485, + "kept_token_count": 736.1265625, + "learning_rate": 1.901946228439179e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9921044528484344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00789554538205266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4490936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7952850341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13290, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.1005750335752964, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.01446036072447896, + "grad_norm": 0.12851455807685852, + "kept_token_count": 735.81875, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.018, + "mgd_cos_sim": 0.992894035577774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007105969451367855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.24358520507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.19732055664065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13300, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10001860335469245, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.015270431619137525, + "grad_norm": 0.12066871672868729, + "kept_token_count": 736.2453125, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9918638110160828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008136186003684997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5620422363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6418762207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13310, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.0997807264328003, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.0185269926674664, + "grad_norm": 0.16008739173412323, + "kept_token_count": 736.8515625, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9924099504947662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007590036559849978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3429443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.72979736328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13320, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.09988115951418877, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.01633428055793047, + "grad_norm": 0.08991118520498276, + "kept_token_count": 736.5609375, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9924517393112182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075482637621462345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.2933685302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0813232421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13330, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.10033060535788536, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.01451736893504858, + "grad_norm": 0.11284390091896057, + "kept_token_count": 736.078125, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9928616464138031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007138354983180761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.83046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.24330139160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13340, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.0996010035276413, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.014168511889874936, + "grad_norm": 0.13375148177146912, + "kept_token_count": 736.3578125, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9929723501205444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007027649506926537, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.49342956542966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7671691894531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13350, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.10021819248795509, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.017646648455411196, + "grad_norm": 0.12716135382652283, + "kept_token_count": 736.3734375, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9925969481468201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007403061911463738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3288116455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3705230712891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13360, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.09965154156088829, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.012668332317844033, + "grad_norm": 0.09617497026920319, + "kept_token_count": 736.728125, + "learning_rate": 1.889186760191666e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9933407783508301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006659225188195705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.19114685058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8899688720703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10077102035284043, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.014429720398038626, + "grad_norm": 0.12430199980735779, + "kept_token_count": 736.2546875, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927450776100158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00725491363555193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.52899169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3907165527344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13380, + "valid_token_count": 818.7625 + }, + { + "actual_token_mask_ratio": 0.1007722795009613, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.014759311825037003, + "grad_norm": 0.15236246585845947, + "kept_token_count": 735.9796875, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9919599056243896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008040107879787683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.8145477294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.65562133789064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.10038909837603568, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.015749478712677956, + "grad_norm": 0.1963750422000885, + "kept_token_count": 736.0953125, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9922269225120545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00777308065444231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7580596923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.688525390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13400, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.09989246353507042, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.0171457695774734, + "grad_norm": 0.12825077772140503, + "kept_token_count": 736.6765625, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9918690919876099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008130904193967581, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4711212158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.25957946777345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13410, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.09972075000405312, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.017198510002344847, + "grad_norm": 0.13214242458343506, + "kept_token_count": 736.790625, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9924799025058746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075200947932899, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.3218627929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.88194885253904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13420, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.10058993399143219, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01686876146122813, + "grad_norm": 0.1538744866847992, + "kept_token_count": 736.2453125, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9933461129665375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653890945017338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8071594238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.66344299316404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13430, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.09962280318140984, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.01588965766131878, + "grad_norm": 0.11322522908449173, + "kept_token_count": 736.7828125, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9926000535488129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073999370448291305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7256286621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3935913085937, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13440, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.10053704306483269, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.014507690444588661, + "grad_norm": 0.1665320098400116, + "kept_token_count": 736.0390625, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.992647135257721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007352857571095228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7658721923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5343475341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13450, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.09945851862430573, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.014303663745522499, + "grad_norm": 0.1145879402756691, + "kept_token_count": 736.7203125, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9930211186408997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006978872418403626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.88287048339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.41883544921876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13460, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09959155842661857, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.015853040385991335, + "grad_norm": 0.12674923241138458, + "kept_token_count": 736.7328125, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9920906186103821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00790939312428236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.94496154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8315460205078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13470, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1001042827963829, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.0157650301232934, + "grad_norm": 0.09417137503623962, + "kept_token_count": 736.4296875, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9927176654338836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072823500260710715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.74205322265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.93944091796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.09990724325180053, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.013694080151617527, + "grad_norm": 0.08388806879520416, + "kept_token_count": 736.4265625, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926920175552368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007308003306388855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8833953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4756713867188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09995992332696915, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.01560324700549245, + "grad_norm": 0.1079065129160881, + "kept_token_count": 736.34375, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9928357422351837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071642562747001644, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7881622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.231640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13500, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09967739656567573, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.017158575542271137, + "grad_norm": 0.10268367826938629, + "kept_token_count": 736.5390625, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921689391136169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007831070851534606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2977264404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4092559814453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13510, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.09946906119585038, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.013375017605721951, + "grad_norm": 0.14462442696094513, + "kept_token_count": 736.8140625, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928543269634247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00714567368850112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.63640747070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8202728271484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13520, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.1001113511621952, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.01434228392317891, + "grad_norm": 0.09463043510913849, + "kept_token_count": 736.35625, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9923279106616973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007672103866934776, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6524169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0276824951172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09996360391378403, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.01540862200781703, + "grad_norm": 0.1200527474284172, + "kept_token_count": 736.18125, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9925978600978851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007402146607637406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.2822326660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.46448974609376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13540, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.10008756667375565, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.016087965061888098, + "grad_norm": 0.0902506485581398, + "kept_token_count": 736.44375, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9930032074451447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006996796000748873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2083374023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.50926818847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13550, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.10003824010491372, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.016708375047892333, + "grad_norm": 0.1179957240819931, + "kept_token_count": 736.4875, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.992453682422638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00754633080214262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9337615966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1195922851563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13560, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09986519813537598, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.016200502682477236, + "grad_norm": 0.07608141005039215, + "kept_token_count": 736.6140625, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.02, + "mgd_cos_sim": 0.992405915260315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007594094146043062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3965362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.75599670410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13570, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09979905784130097, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.015942868031561374, + "grad_norm": 0.10044604539871216, + "kept_token_count": 736.6609375, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925155103206634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007484500017017126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.97430114746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8473693847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13580, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.09993048086762428, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.01701235957443714, + "grad_norm": 0.10292596369981766, + "kept_token_count": 736.3, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9924216628074646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007578340638428926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1118133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.59520568847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13590, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.09968986734747887, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.014995647594332695, + "grad_norm": 0.14048615097999573, + "kept_token_count": 736.5375, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9927907645702362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209237478673458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.09063110351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4162231445313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13600, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.09999183490872383, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.013153795804828405, + "grad_norm": 0.12875355780124664, + "kept_token_count": 736.6046875, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.991960632801056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008039379119873047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.33258056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.42754516601565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13610, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.09992838725447654, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.014923273026943207, + "grad_norm": 0.11311522126197815, + "kept_token_count": 736.328125, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9921996593475342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007800335250794888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.04437561035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.08156433105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13620, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.10011811703443527, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014350417163223028, + "grad_norm": 0.129413902759552, + "kept_token_count": 736.6109375, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9925914287567139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007408557180315256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6628479003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8343872070312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13630, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1004401408135891, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.013655800046399236, + "grad_norm": 0.1235494539141655, + "kept_token_count": 736.23125, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9927932918071747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007206707447767258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9798583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.51763610839845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13640, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.10013717710971833, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.016427235212177038, + "grad_norm": 0.12794806063175201, + "kept_token_count": 736.4265625, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9925948977470398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405110169202089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.709716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.09417419433595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13650, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1004886880517006, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.013821804244071246, + "grad_norm": 0.11885889619588852, + "kept_token_count": 736.1015625, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9928544163703918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071455840021371845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9153259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.74254150390624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.0988458275794983, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.0154661332257092, + "grad_norm": 0.12289261817932129, + "kept_token_count": 737.6875, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.992104297876358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007895706128329039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2273376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2584594726562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13670, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.10038901194930076, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.01879274807870388, + "grad_norm": 0.17873522639274597, + "kept_token_count": 735.9765625, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9924880146980286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007511977758258581, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.2449462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8580657958984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.09971691220998764, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.01576347555965185, + "grad_norm": 0.10654811561107635, + "kept_token_count": 736.6859375, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9919384539127349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0080615415237844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9780731201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7572784423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13690, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.0998154491186142, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.016404492687433957, + "grad_norm": 0.11571308225393295, + "kept_token_count": 736.4875, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9925588965415955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441115006804467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8399932861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.43158264160155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13700, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.09971785172820091, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.016342086903750896, + "grad_norm": 0.1480216532945633, + "kept_token_count": 736.490625, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.992377495765686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00762249194085598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.38408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.64012756347654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.09975827708840371, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.014019101951271295, + "grad_norm": 0.10539250820875168, + "kept_token_count": 736.9734375, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9923455834388732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007654420472681523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.879736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7867462158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.10033128336071968, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.0180948618799448, + "grad_norm": 0.10940737277269363, + "kept_token_count": 735.865625, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9926352798938751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007364722061902284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7566833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6173370361328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13730, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.10005562752485275, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.013955049961805344, + "grad_norm": 0.10747944563627243, + "kept_token_count": 736.6890625, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9932415008544921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006758498400449753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9429595947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7431945800781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13740, + "valid_token_count": 818.59375 + }, + { + "actual_token_mask_ratio": 0.10017147958278656, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.017836039513349534, + "grad_norm": 0.1339637190103531, + "kept_token_count": 736.3921875, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9930633425712585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006936660967767239, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2867736816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6315216064453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13750, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.1000036932528019, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.01665132148191333, + "grad_norm": 0.1233375072479248, + "kept_token_count": 736.3953125, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9920794665813446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00792054207995534, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4709716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.99317626953126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.09977746233344079, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.012869324022904038, + "grad_norm": 0.11506174504756927, + "kept_token_count": 737.0796875, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929336667060852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007066329102963209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.35691833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.2530181884766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13770, + "valid_token_count": 818.7765625 + }, + { + "actual_token_mask_ratio": 0.09927881732583047, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.013540346547961235, + "grad_norm": 0.11863475292921066, + "kept_token_count": 737.121875, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9925987720489502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074012262746691706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8939270019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2467315673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09943345710635185, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.014169103465974331, + "grad_norm": 0.10121715813875198, + "kept_token_count": 736.6671875, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927516341209411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007248371746391058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5849548339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4584564208984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13790, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.09990554675459862, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.013730656076222657, + "grad_norm": 0.10399370640516281, + "kept_token_count": 736.3984375, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9925534427165985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007446566224098205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.19724426269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6484710693359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.09954303354024888, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.01422522533684969, + "grad_norm": 0.10271448642015457, + "kept_token_count": 737.021875, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9922904074192047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007709565106779337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.7456939697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.97374267578124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13810, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.09990668147802353, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.016027892753481866, + "grad_norm": 0.11436973512172699, + "kept_token_count": 736.3875, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9921668946743012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007833103463053704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.08360290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3269409179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13820, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.09986612945795059, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.016995993070304395, + "grad_norm": 0.1114579513669014, + "kept_token_count": 736.6578125, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922177970409394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007782197184860706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.84705505371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3428588867188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13830, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09986695721745491, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.016828566789627075, + "grad_norm": 0.10506846010684967, + "kept_token_count": 736.703125, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.992716658115387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007283351384103298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.39841613769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6110412597656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13840, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.09932769909501075, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015087869018316269, + "grad_norm": 0.12356378883123398, + "kept_token_count": 736.80625, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9923396229743957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007660376373678446, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.58612670898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6096160888672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13850, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.09977221712470055, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.012780304253101348, + "grad_norm": 0.12706072628498077, + "kept_token_count": 736.8265625, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928458333015442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007154185883700847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6636505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8038391113281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13860, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.09983551949262619, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.016201873682439326, + "grad_norm": 0.08924741297960281, + "kept_token_count": 736.7296875, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9926547288894654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007345262914896011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.95318603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.26231079101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13870, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.10054250285029412, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.015879094786942004, + "grad_norm": 0.0948401466012001, + "kept_token_count": 736.203125, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9929153263568878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007084667962044477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6103790283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5659973144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13880, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.09934630990028381, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.01536574074998498, + "grad_norm": 0.10348954051733017, + "kept_token_count": 736.6578125, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9925161957740783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007483817264437675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9009521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.48569946289064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13890, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.09952790439128875, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.014870682079344989, + "grad_norm": 0.10226164758205414, + "kept_token_count": 736.803125, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.992191755771637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007808239478617906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1580383300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.91564025878904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13900, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.09975837916135788, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.014728756621479988, + "grad_norm": 0.14213649928569794, + "kept_token_count": 736.4796875, + "learning_rate": 1.802318716463359e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9933882296085358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006611770950257778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3682342529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6057861328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10051494836807251, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.013663892541080714, + "grad_norm": 0.11250997334718704, + "kept_token_count": 735.846875, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926846146583557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007315390836447477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.87763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2755126953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13920, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.10042611211538315, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.013097193371504546, + "grad_norm": 0.11948098242282867, + "kept_token_count": 735.878125, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9932676792144776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006732320040464401, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0457489013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9727020263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13930, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.0997182659804821, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.013396098930388688, + "grad_norm": 0.12383276224136353, + "kept_token_count": 736.9078125, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9921985566616058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007801440078765154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5375061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7303527832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13940, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.0997539333999157, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.015577960386872291, + "grad_norm": 0.09896611422300339, + "kept_token_count": 736.5109375, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9925645291805267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074354737997055055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.32164306640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.70054626464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13950, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.10035277381539345, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.015064849238842725, + "grad_norm": 0.11922045797109604, + "kept_token_count": 736.2609375, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9930931150913238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906887143850326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8371215820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8569580078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13960, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.09975185841321946, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.01711603030562401, + "grad_norm": 0.10184197872877121, + "kept_token_count": 736.3921875, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9930492281913758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006950768176466227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.64008178710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3860778808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13970, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.1006829246878624, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.013681965414434671, + "grad_norm": 0.10817117989063263, + "kept_token_count": 735.534375, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930791258811951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006920884922146797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2687683105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.11827697753904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13980, + "valid_token_count": 817.8828125 + }, + { + "actual_token_mask_ratio": 0.10085751116275787, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.014696767833083869, + "grad_norm": 0.11431171000003815, + "kept_token_count": 735.5046875, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926126360893249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007387369219213724, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.36378784179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.23801879882814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 13990, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.10018715634942055, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.012771643977612258, + "grad_norm": 0.08562058955430984, + "kept_token_count": 736.396875, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9923275113105774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007672494370490312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4090087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1418212890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14000, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.100498166680336, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.015050723683089018, + "grad_norm": 0.1324746161699295, + "kept_token_count": 735.534375, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927917361259461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007208265457302332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8314666748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.07838439941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14010, + "valid_token_count": 817.7140625 + }, + { + "actual_token_mask_ratio": 0.10002667084336281, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.015958374831825495, + "grad_norm": 0.09384109824895859, + "kept_token_count": 736.4328125, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9927673220634461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007232682220637799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.991455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.10596923828126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14020, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.09922882169485092, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.013410014612600208, + "grad_norm": 0.07129350304603577, + "kept_token_count": 736.9421875, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.992697149515152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007302862871438265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6621917724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.51520385742185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14030, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10037209093570709, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.012463752180337906, + "grad_norm": 0.08591704815626144, + "kept_token_count": 736.128125, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9928099155426026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00719008194282651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.81089477539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7075653076172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14040, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09991898536682128, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.017649095691740512, + "grad_norm": 0.11788526177406311, + "kept_token_count": 736.640625, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9925519108772278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074480828829109665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1878326416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7110626220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14050, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.1005401112139225, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.015319056436419487, + "grad_norm": 0.1076270341873169, + "kept_token_count": 735.8671875, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9919993281364441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008000664785504341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.39552917480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6565002441406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14060, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10034165754914284, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.011848113872110844, + "grad_norm": 0.10252822935581207, + "kept_token_count": 736.678125, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.992527449131012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007472558505833149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3001007080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.48162536621095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14070, + "valid_token_count": 818.8453125 + }, + { + "actual_token_mask_ratio": 0.10024026855826378, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.013784253783524037, + "grad_norm": 0.09958260506391525, + "kept_token_count": 736.1390625, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.992268443107605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007731552608311177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.655615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4617156982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14080, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.10070003196597099, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.014016016013920307, + "grad_norm": 0.09281209111213684, + "kept_token_count": 735.5171875, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9921481549739838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00785183971747756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.34987487792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.2134521484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14090, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.1001213513314724, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.01389712980017066, + "grad_norm": 0.09802736341953278, + "kept_token_count": 736.4046875, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9925823092460633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007417692895978689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8752868652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.20115966796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14100, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.09988085478544236, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.014345775544643401, + "grad_norm": 0.14452144503593445, + "kept_token_count": 736.19375, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9926816821098328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00731832729652524, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.78076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.04905700683594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14110, + "valid_token_count": 817.8859375 + }, + { + "actual_token_mask_ratio": 0.09993129447102547, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.015913556888699533, + "grad_norm": 0.10243906080722809, + "kept_token_count": 736.5609375, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9923977196216583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00760227981954813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8112823486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6049560546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14120, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09944955855607987, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.014733713492751122, + "grad_norm": 0.08471866697072983, + "kept_token_count": 736.871875, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9929066002368927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00709340525791049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.582177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5554992675781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14130, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.10000100582838059, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.013057100307196378, + "grad_norm": 0.10042204707860947, + "kept_token_count": 736.471875, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9931238710880279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068761340342462065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5124176025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.41107788085935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14140, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.09998421296477318, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.012713748961687088, + "grad_norm": 0.13329611718654633, + "kept_token_count": 736.7890625, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9929817616939545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007018227223306894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2224853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8096588134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14150, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.09988654777407646, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.014830880519002675, + "grad_norm": 0.11569953709840775, + "kept_token_count": 736.484375, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9921637773513794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007836222741752863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.15776977539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.59910583496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14160, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.09952829927206039, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.014719558600336313, + "grad_norm": 0.12902146577835083, + "kept_token_count": 736.9015625, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9930266439914703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006973363179713488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7558837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.84543762207034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1003596805036068, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.014635648671537637, + "grad_norm": 0.12767884135246277, + "kept_token_count": 736.1640625, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926569759845734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007343009114265442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.6512847900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.53269348144534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14180, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09994491413235665, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.015447932295501232, + "grad_norm": 0.12515783309936523, + "kept_token_count": 736.453125, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9923531174659729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007646876014769077, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6089172363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7606475830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.10020011365413665, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.01497603552415967, + "grad_norm": 0.09695886820554733, + "kept_token_count": 736.2078125, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9927383720874786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261620368808508, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.24681701660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.32115478515624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.1002245657145977, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.015622699074447155, + "grad_norm": 0.12577718496322632, + "kept_token_count": 736.4078125, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922522008419037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007747802510857582, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.32984313964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.76954345703126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14210, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09945942834019661, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.016689302399754524, + "grad_norm": 0.10726048797369003, + "kept_token_count": 736.64375, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9926962912082672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007303709629923105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5573211669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4086639404297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14220, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.1006740115582943, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.015591112896800041, + "grad_norm": 0.14897257089614868, + "kept_token_count": 735.9765625, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927813589572907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007218640949577093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.41733703613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7875701904297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14230, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.0995133712887764, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.014433028735220432, + "grad_norm": 0.12796452641487122, + "kept_token_count": 736.521875, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927557587623597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007244253810495138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.0923126220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.09113159179685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14240, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.09987259656190872, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.01645739087834954, + "grad_norm": 0.10654928535223007, + "kept_token_count": 736.303125, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9923748016357422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076251918449997905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1259521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.21002502441405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.1004326656460762, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.014033287577331067, + "grad_norm": 0.096815325319767, + "kept_token_count": 736.1984375, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9923895120620727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076105033978819845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.4905944824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.05702209472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14260, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10073829665780068, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.01626883614808321, + "grad_norm": 0.14124010503292084, + "kept_token_count": 735.7609375, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9929432153701783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007056789007037878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.0957550048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2660797119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14270, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.10007927194237709, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.014508222229778767, + "grad_norm": 0.1451234668493271, + "kept_token_count": 736.603125, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9920071184635162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007992884609848261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.43617248535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0406158447266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14280, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.10099992230534553, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.01395518439821899, + "grad_norm": 0.09040969610214233, + "kept_token_count": 735.6875, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.993020212650299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006979785300791264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.34974670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2373046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.1001010812819004, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.01601659655570984, + "grad_norm": 0.08892112970352173, + "kept_token_count": 736.35625, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925640642642974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007435938715934754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3927825927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5029357910156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14300, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09958138018846512, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.017909943219274284, + "grad_norm": 0.08853700757026672, + "kept_token_count": 736.65625, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.992396068572998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076039302162826065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.80925903320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.39654235839845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14310, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.09983162358403205, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.015187882352620363, + "grad_norm": 0.13348844647407532, + "kept_token_count": 736.7703125, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9929267108440399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007073273696005344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.03868103027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.95513305664065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14320, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.09969466403126717, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.013534688018262386, + "grad_norm": 0.12276137620210648, + "kept_token_count": 736.7359375, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9920943200588226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007905673887580632, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9306121826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.64012756347654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14330, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.09946120753884316, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.01771061746403575, + "grad_norm": 0.07559895515441895, + "kept_token_count": 736.5484375, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9929166972637177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007083305716514587, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0846923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.4804351806641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14340, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.09990663900971412, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.012786285066977144, + "grad_norm": 0.09478051960468292, + "kept_token_count": 736.4765625, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.993079423904419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00692057004198432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7278656005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0234619140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.10009258091449738, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.014802446495741606, + "grad_norm": 0.08868464082479477, + "kept_token_count": 736.278125, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9924908876419067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007509116642177105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3158477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.55335998535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14360, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.0997236005961895, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.013252425519749523, + "grad_norm": 0.08237951993942261, + "kept_token_count": 737.0296875, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992679488658905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007320507429540157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.82860107421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1887634277344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14370, + "valid_token_count": 818.6671875 + }, + { + "actual_token_mask_ratio": 0.10037645548582078, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.014615837670862674, + "grad_norm": 0.09695485234260559, + "kept_token_count": 736.4140625, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928959846496582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007104016840457916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.93565979003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.31952514648435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14380, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.10045400634407997, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.015912579838186502, + "grad_norm": 0.10231677442789078, + "kept_token_count": 735.8890625, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9922964990139007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077035176567733284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.3517822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6007049560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14390, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.09957867264747619, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.013748234044760466, + "grad_norm": 0.10543052852153778, + "kept_token_count": 736.7265625, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9929168581962585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007083133328706026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.54864501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.24757080078126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14400, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.1001335285604, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.013716026954352856, + "grad_norm": 0.1065383180975914, + "kept_token_count": 736.2234375, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926823437213897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007317650504410267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6615325927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1660888671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14410, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.10006232932209969, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.012711381958797573, + "grad_norm": 0.07746266573667526, + "kept_token_count": 736.2265625, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9927068471908569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007293171714991331, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.7383087158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.73000183105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14420, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.10029591396450996, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.014431573171168565, + "grad_norm": 0.10974902659654617, + "kept_token_count": 735.96875, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9920805752277374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007919423840939998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2984558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1676055908203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14430, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.10030245706439019, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.016293786000460386, + "grad_norm": 0.12686757743358612, + "kept_token_count": 736.1671875, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9930338978767395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006966097373515367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5464202880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5216033935547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14440, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.09966303035616875, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.013034301716834307, + "grad_norm": 0.1286177933216095, + "kept_token_count": 736.834375, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9932592988014222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006740700174123049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.24070129394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9342041015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14450, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.09949839562177658, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.013624979835003614, + "grad_norm": 0.11375318467617035, + "kept_token_count": 736.4828125, + "learning_rate": 1.712743183030592e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9931523203849792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006847670394927263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1063720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.69300537109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14460, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.09945653975009919, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.01383205372840166, + "grad_norm": 0.10436299443244934, + "kept_token_count": 737.0953125, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.992794668674469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00720533262938261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2196533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.92564392089844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14470, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10048632547259331, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.012507396936416625, + "grad_norm": 0.11773332953453064, + "kept_token_count": 736.315625, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9928713917732239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007128605712205171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.3819580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5542022705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14480, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.09966148659586907, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.014449586253613233, + "grad_norm": 0.07318263500928879, + "kept_token_count": 736.65, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9923029243946075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007697095070034266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.12481079101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.45948181152346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14490, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10007050409913062, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.013552323542535305, + "grad_norm": 0.07528412342071533, + "kept_token_count": 736.575, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9927319049835205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007268084771931171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5015411376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.55816040039065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14500, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.0993303969502449, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.016366418451070786, + "grad_norm": 0.11620213091373444, + "kept_token_count": 737.0078125, + "learning_rate": 1.704556217045536e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9927202939987183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007279712148010731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.88854064941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.86912536621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14510, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09941707029938698, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.014507268462330103, + "grad_norm": 0.12326447665691376, + "kept_token_count": 737.1359375, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.992744380235672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007255610078573227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.24896545410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3779663085937, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14520, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.09908157736063003, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.014761085715144872, + "grad_norm": 0.09963546693325043, + "kept_token_count": 736.775, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9927225470542907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007277448102831841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.23829345703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.09361877441404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14530, + "valid_token_count": 817.80625 + }, + { + "actual_token_mask_ratio": 0.10041726231575013, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.01344596380367875, + "grad_norm": 0.10574492067098618, + "kept_token_count": 736.1296875, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930196762084961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006980323605239392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.76755981445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.1820068359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14540, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.0998144045472145, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.015564493276178837, + "grad_norm": 0.11042360961437225, + "kept_token_count": 736.578125, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9926958799362182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007304107025265694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.5892791748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.54071655273435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14550, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.0995805449783802, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.015027867536991835, + "grad_norm": 0.10461653023958206, + "kept_token_count": 736.6546875, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9929089903831482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007091000117361546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5049713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2858612060547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14560, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.10010962933301926, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.014043353497982025, + "grad_norm": 0.0802498608827591, + "kept_token_count": 736.3390625, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926211357116699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073788645677268505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4059600830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.44556579589846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14570, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.09996499195694923, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.014013293152675032, + "grad_norm": 0.09603988379240036, + "kept_token_count": 736.4390625, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9925760746002197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007423924934118986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0028015136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1587158203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14580, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10041639059782029, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.014170912373811007, + "grad_norm": 0.15139465034008026, + "kept_token_count": 736.08125, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927825391292572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007217450998723507, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.29314575195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3438659667969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14590, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.10053621158003807, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.013456403696909546, + "grad_norm": 0.07723306119441986, + "kept_token_count": 735.9734375, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9933859646320343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066140260547399524, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3870513916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4854370117188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14600, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.0990568906068802, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.014811050519347191, + "grad_norm": 0.09004829823970795, + "kept_token_count": 736.9734375, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9927258014678955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007274201884865761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8306365966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7946044921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14610, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.10008802264928818, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.013601171877235174, + "grad_norm": 0.1295837163925171, + "kept_token_count": 736.2125, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9925306677818299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007469323556870222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0393035888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2958465576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14620, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.09956910684704781, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.01357489456422627, + "grad_norm": 0.08985797315835953, + "kept_token_count": 736.8453125, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930250465869903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006974969059228897, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4384490966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.70368041992185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.10091350823640824, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.014433478657156228, + "grad_norm": 0.11755529046058655, + "kept_token_count": 735.6375, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9930272042751312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006972798705101013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.29717407226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3301239013672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14640, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.10025949403643608, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.013607416953891515, + "grad_norm": 0.1231178417801857, + "kept_token_count": 736.3203125, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9923188269138337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007681159861385822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3366333007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.61499938964846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09960899204015732, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.013990650232881308, + "grad_norm": 0.11816089600324631, + "kept_token_count": 736.8140625, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.99249769449234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007502305693924427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.45409240722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.47313232421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14660, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.09933008998632431, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.017232402134686707, + "grad_norm": 0.10007665306329727, + "kept_token_count": 737.109375, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9922905921936035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077094055712223055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3227142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9601623535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14670, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.09942967817187309, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.01547740874812007, + "grad_norm": 0.10879667848348618, + "kept_token_count": 737.165625, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9923731207847595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007626862544566393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.48841552734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.505029296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14680, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.099714095890522, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.013421132788062095, + "grad_norm": 0.10030799359083176, + "kept_token_count": 736.6234375, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.993031233549118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006968752667307854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.65330505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7144012451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14690, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.0996100775897503, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.01629279972985387, + "grad_norm": 0.11628992110490799, + "kept_token_count": 736.8984375, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9929715991020203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007028409745544195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6442352294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.8506195068359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14700, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.09980417490005493, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.013181889802217484, + "grad_norm": 0.09707280248403549, + "kept_token_count": 736.8078125, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926727771759033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007327223476022482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.85728454589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.4341766357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14710, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.10044382587075233, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.014898113254457713, + "grad_norm": 0.12776711583137512, + "kept_token_count": 736.503125, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9925107002258301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007489304803311825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.40028076171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.231591796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14720, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.10080998837947845, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.016964077856391667, + "grad_norm": 0.13818733394145966, + "kept_token_count": 735.8765625, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9927925348281861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007207465823739767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.03807373046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7993682861328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14730, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.10024923980236053, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.013692723819985986, + "grad_norm": 0.10559603571891785, + "kept_token_count": 735.9953125, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9930604279041291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939568463712931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.12717590332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.862744140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14740, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.10035018473863602, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.013192881271243096, + "grad_norm": 0.11440333724021912, + "kept_token_count": 735.965625, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929832875728607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070167209953069685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2628204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8906311035156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14750, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.10072143971920014, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.015554803889244795, + "grad_norm": 0.11992528289556503, + "kept_token_count": 736.2640625, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9929499208927155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070500772446393965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.17950134277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.79412536621095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14760, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.1002827413380146, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.014069997239857912, + "grad_norm": 0.09668175131082535, + "kept_token_count": 736.6015625, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.993315452337265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006684555765241384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.7015838623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7422637939453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14770, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.10023571252822876, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.014391922112554311, + "grad_norm": 0.10718114674091339, + "kept_token_count": 736.1921875, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9926565408706665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007343460526317358, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1828216552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.92799682617186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14780, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.10042268335819245, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.013950111716985703, + "grad_norm": 0.1528952270746231, + "kept_token_count": 736.0890625, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9924632787704468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007536700647324324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.58819885253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0114288330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14790, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.10038988962769509, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.01515594432130456, + "grad_norm": 0.09762841463088989, + "kept_token_count": 735.99375, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9931747078895569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068252976052463055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1488739013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.46677856445314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14800, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09987067058682442, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.015052298456430436, + "grad_norm": 0.10203783214092255, + "kept_token_count": 736.846875, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927421808242798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007257804274559021, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0938354492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.81678466796876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14810, + "valid_token_count": 818.5984375 + }, + { + "actual_token_mask_ratio": 0.10062090009450912, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.012345384154468775, + "grad_norm": 0.0888201966881752, + "kept_token_count": 736.1375, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9930563986301422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006943618040531873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9700622558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9913024902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14820, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.1004154123365879, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.01600311426445842, + "grad_norm": 0.09470982849597931, + "kept_token_count": 736.2375, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920163571834564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007983631547540426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.8434112548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0261169433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14830, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.09932871535420418, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.014742628298699855, + "grad_norm": 0.11036566644906998, + "kept_token_count": 736.7953125, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928586542606354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007141350489109755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7229248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6420623779297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14840, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.1000331163406372, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.012168287206441164, + "grad_norm": 0.07879170030355453, + "kept_token_count": 736.40625, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9927612543106079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007238742895424366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5378173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5826812744141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14850, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10008284151554107, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.014893674291670322, + "grad_norm": 0.10160892456769943, + "kept_token_count": 736.5, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927591860294342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072408108040690426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.7326232910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6732971191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14860, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.099994695186615, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.014781889319419862, + "grad_norm": 0.10607698559761047, + "kept_token_count": 736.490625, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9925460517406464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074539373628795145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.96522827148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.12744140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14870, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.10010746493935585, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.01592792198061943, + "grad_norm": 0.12331315129995346, + "kept_token_count": 736.3046875, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.992563259601593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007436741050332785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2890991210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.22046813964846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14880, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10003282278776168, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.014410962630063296, + "grad_norm": 0.10412710905075073, + "kept_token_count": 736.6234375, + "learning_rate": 1.642150118027194e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927988469600677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007201151642948389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.49583435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7921600341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14890, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.09975005239248276, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.016948386467993258, + "grad_norm": 0.14599967002868652, + "kept_token_count": 736.8578125, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9927670121192932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00723298192024231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.77927856445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7725402832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14900, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.10026530921459198, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.015652471501380206, + "grad_norm": 0.12298765778541565, + "kept_token_count": 736.084375, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.992759519815445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00724048176780343, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.33147583007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8263885498047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.099896290153265, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.013849659916013479, + "grad_norm": 0.09143620729446411, + "kept_token_count": 736.5125, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9931223869323731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006877622753381729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.08409118652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5719268798828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14920, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09982405081391335, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.01673551378771663, + "grad_norm": 0.14968927204608917, + "kept_token_count": 736.5921875, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9926318049430847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007368186954408884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7607818603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8199035644531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14930, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.09980605095624924, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.015498515497893096, + "grad_norm": 0.092452272772789, + "kept_token_count": 736.7078125, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9928124010562897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187607325613498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.64252014160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.84222412109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14940, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09979382678866386, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.013979779463261366, + "grad_norm": 0.12330644577741623, + "kept_token_count": 736.63125, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9925947070121766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405283208936453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.95879516601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2163330078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14950, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.10013183578848839, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.012507376307621599, + "grad_norm": 0.08084945380687714, + "kept_token_count": 736.575, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9930533826351166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006946620438247919, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4790435791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.64873962402345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14960, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.09987687021493911, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.014596738666296006, + "grad_norm": 0.09235981851816177, + "kept_token_count": 736.9875, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928029656410218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007197039853781462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8150360107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.15078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14970, + "valid_token_count": 818.7609375 + }, + { + "actual_token_mask_ratio": 0.09984277263283729, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.01465907422825694, + "grad_norm": 0.08432797342538834, + "kept_token_count": 736.346875, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9932855188846588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006714467518031597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.40498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1994567871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14980, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.09891455620527267, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.013965290877968073, + "grad_norm": 0.12858547270298004, + "kept_token_count": 737.28125, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9931438863277435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006856117490679025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9467529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2413330078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 14990, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10095356777310371, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.014518069848418236, + "grad_norm": 0.11716436594724655, + "kept_token_count": 735.7671875, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9930499970912934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069500069133937355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1709777832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.61098022460936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15000, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.10048322826623916, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.016285207215696575, + "grad_norm": 0.11681752651929855, + "kept_token_count": 736.253125, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9930313289165497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006968680210411549, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6868621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.25165405273435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15010, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.09991996735334396, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.015538846235722303, + "grad_norm": 0.12448494136333466, + "kept_token_count": 736.6125, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9931382238864899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006861772015690803, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.57798156738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.64215698242185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15020, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.09954118505120277, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.013159008510410786, + "grad_norm": 0.11820610612630844, + "kept_token_count": 736.86875, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9928479671478272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007152029126882553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.3924194335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9216003417969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15030, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.09983687773346901, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.013280513416975736, + "grad_norm": 0.10404819250106812, + "kept_token_count": 736.65625, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928384065628052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007161583378911018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.60080871582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.55836486816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.0999392181634903, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.013928261306136846, + "grad_norm": 0.11308962851762772, + "kept_token_count": 736.6515625, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9932897329330445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710274424403906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.35303649902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8663818359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15050, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.09947373047471046, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.014111259719356894, + "grad_norm": 0.09080913662910461, + "kept_token_count": 737.03125, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9937340497970581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006265942100435495, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7906127929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.514208984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15060, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.10017828792333602, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.015466559771448373, + "grad_norm": 0.11986857652664185, + "kept_token_count": 736.103125, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9932014167308807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006798575539141894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0350646972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8733856201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15070, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.09937838912010193, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.017214820999652148, + "grad_norm": 0.131290465593338, + "kept_token_count": 737.0546875, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9928512930870056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007148715760558843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.59778747558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.48752746582034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15080, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.09981557130813598, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.016566680185496806, + "grad_norm": 0.13039878010749817, + "kept_token_count": 736.4640625, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9928985953330993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007101416680961847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6483093261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.46941833496095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15090, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09989176765084266, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.014927961025387048, + "grad_norm": 0.11372917145490646, + "kept_token_count": 736.4078125, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9928237140178681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00717628700658679, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0316650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8740173339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15100, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.10000045821070672, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.01585768428631127, + "grad_norm": 0.09820310771465302, + "kept_token_count": 736.215625, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9926938831806182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007306107226759196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.760595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.04302978515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15110, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.09998070448637009, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.014592231716960669, + "grad_norm": 0.14428210258483887, + "kept_token_count": 736.665625, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9931863248348236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006813694071024656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0210662841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.21978149414065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15120, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.0998916320502758, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.012672686856240034, + "grad_norm": 0.10189509391784668, + "kept_token_count": 736.7859375, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928412139415741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007158797793090343, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2507263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.95042419433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15130, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.10040934085845947, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.015820366609841586, + "grad_norm": 0.152662456035614, + "kept_token_count": 736.30625, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9924660444259643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007533963862806558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.31052856445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.42267456054685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15140, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.09928717538714409, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.01350785857066512, + "grad_norm": 0.09105567634105682, + "kept_token_count": 736.7953125, + "learning_rate": 1.599300861357545e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9930800199508667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00691996244713664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6754547119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3758178710938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15150, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.1001083105802536, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.013891442772001028, + "grad_norm": 0.1350039541721344, + "kept_token_count": 736.0796875, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9922568798065186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007743137329816819, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.0069061279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.24061584472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15160, + "valid_token_count": 817.965625 + }, + { + "actual_token_mask_ratio": 0.1000187337398529, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.015150722302496433, + "grad_norm": 0.08657490462064743, + "kept_token_count": 736.2671875, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9925542950630188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007445702236145735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0982971191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8805236816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15170, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10004200190305709, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.014930373709648847, + "grad_norm": 0.11559417098760605, + "kept_token_count": 736.6, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9928647994995117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007135205529630184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.11745300292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3983489990234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15180, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.09971014112234115, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.01693744333460927, + "grad_norm": 0.1108333095908165, + "kept_token_count": 736.646875, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9925614893436432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007438511028885842, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9063415527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.76083984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.10022518932819366, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.01482615526765585, + "grad_norm": 0.10162937641143799, + "kept_token_count": 736.0703125, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9930012047290802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00699878828600049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0005645751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.39418640136716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15200, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.09944573119282722, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.013597448449581861, + "grad_norm": 0.10218861699104309, + "kept_token_count": 737.0609375, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930562853813172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069437045603990555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9634948730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0321563720703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15210, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.10009892582893372, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.014410959091037512, + "grad_norm": 0.09292808175086975, + "kept_token_count": 736.1140625, + "learning_rate": 1.587749121184485e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927639424800873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072360456921160225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.88153686523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5047912597656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15220, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.09989456832408905, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.01530541544780135, + "grad_norm": 0.11034801602363586, + "kept_token_count": 736.5390625, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.992775845527649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224157638847828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7354248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2303131103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15230, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.10015152096748352, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.015465012192726136, + "grad_norm": 0.13379508256912231, + "kept_token_count": 736.425, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925755620002746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074244382791221145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.45411071777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.40123291015624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.10025925412774087, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.014448402635753154, + "grad_norm": 0.12756463885307312, + "kept_token_count": 736.0046875, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9928143739700317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071856121532619, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4665191650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.38136291503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15250, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.09953751489520073, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.01359382076188922, + "grad_norm": 0.10158797353506088, + "kept_token_count": 736.715625, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9928093254566193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007190680131316185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7364135742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.7258758544922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15260, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.10068109259009361, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.017766780965030194, + "grad_norm": 0.086968794465065, + "kept_token_count": 735.81875, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9928201198577881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007179864123463631, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6029846191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.68901977539065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15270, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.10065686777234077, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.015234522148966789, + "grad_norm": 0.08475133031606674, + "kept_token_count": 736.0109375, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9932022511959075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006797764264047146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.94496154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9096984863281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15280, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.10042567774653435, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.013898377120494843, + "grad_norm": 0.10013162344694138, + "kept_token_count": 736.0453125, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9932206153869629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006779385730624199, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.14362182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7296966552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15290, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.10013574436306953, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.015355557342991233, + "grad_norm": 0.12866650521755219, + "kept_token_count": 736.346875, + "learning_rate": 1.574540775157511e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926152110099793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007384786754846573, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9900848388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8266235351563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15300, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.10052714943885803, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.0133495828602463, + "grad_norm": 0.09869422763586044, + "kept_token_count": 736.034375, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9924259901046752, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007574012409895659, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.2998504638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3498291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15310, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.09991011470556259, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.012026585172861814, + "grad_norm": 0.07744382321834564, + "kept_token_count": 736.5953125, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9924286067485809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075713864527642725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5045166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.74102172851565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15320, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09954161569476128, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.014151597395539284, + "grad_norm": 0.09092896431684494, + "kept_token_count": 736.553125, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9931191980838776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006880819890648126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.24314575195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4405548095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15330, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.10042237713932992, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.012788997031748295, + "grad_norm": 0.10449773073196411, + "kept_token_count": 736.084375, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9932606995105744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067392983473837376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8726318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8641845703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15340, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.10000104829668999, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.014496844774112106, + "grad_norm": 0.1001674234867096, + "kept_token_count": 736.346875, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9931688785552979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068311383947730064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8455108642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2556854248047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15350, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.09992213770747185, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.01506873993203044, + "grad_norm": 0.09896378964185715, + "kept_token_count": 736.7671875, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9930737435817718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069262546487152575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.07785949707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4099395751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15360, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.1002491518855095, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.016449104715138673, + "grad_norm": 0.10627192258834839, + "kept_token_count": 736.3375, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9926347017288208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007365297246724367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.8630615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8138732910156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15370, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.09940332695841789, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.013873045984655618, + "grad_norm": 0.10721146315336227, + "kept_token_count": 736.821875, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992642593383789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007357394602149725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.92083740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.44700317382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15380, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.09978149831295013, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.015830970462411643, + "grad_norm": 0.12031742185354233, + "kept_token_count": 736.60625, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9928947269916535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007105271890759468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.90330200195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.984228515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15390, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.1005462646484375, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.014212800469249488, + "grad_norm": 0.10067626088857651, + "kept_token_count": 736.2703125, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9929356098175048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070643887855112554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6750762939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4593200683594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15400, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.09975502118468285, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.012325627077370883, + "grad_norm": 0.10326658934354782, + "kept_token_count": 736.6015625, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9928470551967621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007152951881289482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1559173583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.59837341308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15410, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09952561259269714, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.01644980553537607, + "grad_norm": 0.11589932441711426, + "kept_token_count": 736.9515625, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9928836047649383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007116395980119705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.064208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.15074768066404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15420, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.09995180368423462, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.013810168951749802, + "grad_norm": 0.11500710994005203, + "kept_token_count": 736.5703125, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.992594736814499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405262067914009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1314178466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3068420410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15430, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.09995167255401612, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.016055004112422465, + "grad_norm": 0.11511797457933426, + "kept_token_count": 736.6421875, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9925981521606445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074018527753651146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8778411865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9834381103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15440, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.09987324103713036, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.013317391090095044, + "grad_norm": 0.09736976027488708, + "kept_token_count": 736.6703125, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9924431324005127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007556889299303293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.4678100585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.84288330078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15450, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.09952492043375968, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.014108875766396523, + "grad_norm": 0.10069243609905243, + "kept_token_count": 736.6953125, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9925389885902405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074610174633562565, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7918243408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.40379638671874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15460, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.09950499013066291, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.011828259471803904, + "grad_norm": 0.08607852458953857, + "kept_token_count": 736.9203125, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9927835404872895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007216466404497623, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.61403503417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.86590576171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.09983977749943733, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.015611171163618565, + "grad_norm": 0.09205455332994461, + "kept_token_count": 736.0703125, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927243113517761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072756870649755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3563171386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7206573486328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15480, + "valid_token_count": 817.709375 + }, + { + "actual_token_mask_ratio": 0.1003914624452591, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.014878853969275952, + "grad_norm": 0.09048958122730255, + "kept_token_count": 735.996875, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.992801034450531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007198967784643173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7806396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.75522766113284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15490, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.1002354733645916, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.01589973960071802, + "grad_norm": 0.09683369845151901, + "kept_token_count": 736.1203125, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9926103949546814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073896004818379876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.67601318359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5269317626953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15500, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.09968746900558471, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.01352823581546545, + "grad_norm": 0.10204727947711945, + "kept_token_count": 736.909375, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926423132419586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007357679028064013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.1833129882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8921264648437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15510, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.0992192193865776, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.015721233328804374, + "grad_norm": 0.1374509185552597, + "kept_token_count": 736.921875, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9928590893745423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007140905596315861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.53284912109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.94480590820314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15520, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.09981206953525543, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.012863418972119688, + "grad_norm": 0.13529323041439056, + "kept_token_count": 736.4734375, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9935547888278962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006445206888020039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2273681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.56128845214846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15530, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.09995267540216446, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.014379280898720027, + "grad_norm": 0.12264153361320496, + "kept_token_count": 736.0875, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.992399924993515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007600074354559183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0897644042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.41400146484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15540, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.09942320212721825, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.015536079742014408, + "grad_norm": 0.08940067887306213, + "kept_token_count": 737.2203125, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.992813789844513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007186211366206408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.59447021484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.00474243164064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15550, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.10021360814571381, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.01555071510374546, + "grad_norm": 0.10347461700439453, + "kept_token_count": 736.0984375, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9929574251174926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007042581308633089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.28818969726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.49578857421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15560, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.09947700574994087, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.016605121456086634, + "grad_norm": 0.12009773403406143, + "kept_token_count": 736.9, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9925442516803742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007455734629184007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.3840026855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.23024597167966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15570, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.10031276419758797, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.015382702928036451, + "grad_norm": 0.08384682983160019, + "kept_token_count": 736.68125, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9929898738861084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007010109722614288, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.6458404541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7906005859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15580, + "valid_token_count": 818.815625 + }, + { + "actual_token_mask_ratio": 0.10054628849029541, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.013512490782886744, + "grad_norm": 0.10948467254638672, + "kept_token_count": 736.059375, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926169455051422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007383057381957769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.73850708007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9048583984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15590, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.0998599350452423, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.01562890103086829, + "grad_norm": 0.09723677486181259, + "kept_token_count": 736.8828125, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922550797462464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007744908146560192, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9668884277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3860168457031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15600, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.1003745898604393, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.014631985919550061, + "grad_norm": 0.13474448025226593, + "kept_token_count": 736.4875, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992692619562149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007307388167828321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.81285095214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0437805175781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15610, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.09950125217437744, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.012979530356824398, + "grad_norm": 0.11567022651433945, + "kept_token_count": 736.8828125, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9924413919448852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007558613270521164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.73227844238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.20838317871096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15620, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.100318044424057, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.013043570797890425, + "grad_norm": 0.10449977964162827, + "kept_token_count": 736.2203125, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9928792119026184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007120789028704166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.62059631347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.62453308105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15630, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.10054688304662704, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.01482567898929119, + "grad_norm": 0.10238026082515717, + "kept_token_count": 736.209375, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9925018906593323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007498104404658079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.8765594482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6601837158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15640, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.10002497881650925, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.013339087832719088, + "grad_norm": 0.10623341798782349, + "kept_token_count": 736.4140625, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992864066362381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071359344758093355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.86228942871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.09539184570315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15650, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.09931089356541634, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.015203169733285903, + "grad_norm": 0.11096059530973434, + "kept_token_count": 737.2453125, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9927032589912415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072967388667166235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2867370605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3588043212891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15660, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.10059425905346871, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.012797667365521192, + "grad_norm": 0.09888315200805664, + "kept_token_count": 736.1890625, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9926794648170472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00732054328545928, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.06578674316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7668823242187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15670, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.10022049844264984, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.013294747704640031, + "grad_norm": 0.1114892065525055, + "kept_token_count": 736.221875, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9927997350692749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007200259622186423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.37952880859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.33802795410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09973385855555535, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.012768283486366272, + "grad_norm": 0.12042193114757538, + "kept_token_count": 736.6234375, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930318355560303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069681589491665365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9751312255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.20855712890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15690, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.10043555125594139, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.013926399778574705, + "grad_norm": 0.09516259282827377, + "kept_token_count": 735.934375, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9923262476921082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007673750910907984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.87547912597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.92682495117185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15700, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.10047277063131332, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.014416045555844904, + "grad_norm": 0.10277853161096573, + "kept_token_count": 735.79375, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927365720272064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007263424806296825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4195190429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.34790954589846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15710, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.09957350865006447, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013186992518603802, + "grad_norm": 0.11900730431079865, + "kept_token_count": 736.85625, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930427312850952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006957269739359617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.03191528320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1088623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09925737157464028, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.013968168664723635, + "grad_norm": 0.1045333668589592, + "kept_token_count": 737.115625, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9926695168018341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007330479938536883, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.9907196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7216064453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15730, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.09964236617088318, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.013367534475401044, + "grad_norm": 0.13205480575561523, + "kept_token_count": 736.5796875, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992874026298523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00712596969678998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.22430114746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.40269470214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15740, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.09959222301840782, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.015026176627725363, + "grad_norm": 0.09508522599935532, + "kept_token_count": 736.484375, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9931756496429444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006824361719191074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.84413146972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3994476318359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15750, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.09970713183283805, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.014281364530324936, + "grad_norm": 0.1448293775320053, + "kept_token_count": 736.6234375, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9923267364501953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007673269230872392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.75697021484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.66644897460935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15760, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.10024606883525848, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.014959027711302041, + "grad_norm": 0.10663188248872757, + "kept_token_count": 736.1703125, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9932921946048736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006707809399813414, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.01320495605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.42546997070315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.10056399777531624, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.014863903168588877, + "grad_norm": 0.12883874773979187, + "kept_token_count": 735.7078125, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927234530448914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007276543229818344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.39008178710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7024230957031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15780, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.09974252730607987, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.014807012211531401, + "grad_norm": 0.10738880932331085, + "kept_token_count": 736.63125, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9932684540748596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006731531862169504, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7404418945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.00584411621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10039457902312279, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.015401224698871374, + "grad_norm": 0.12139639258384705, + "kept_token_count": 736.38125, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9928732812404633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007126734126359224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.755029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.01707458496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15800, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.10063390731811524, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.01429126812145114, + "grad_norm": 0.11083772778511047, + "kept_token_count": 735.746875, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927051782608032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007294837571680546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.54909362792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6896148681641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15810, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.10008183643221855, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.014302392490208149, + "grad_norm": 0.1421487182378769, + "kept_token_count": 736.2515625, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927369832992554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072630058974027635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7077667236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8135009765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15820, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.10003260746598244, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.013064130442216992, + "grad_norm": 0.127817764878273, + "kept_token_count": 736.4359375, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9934595346450805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00654045520350337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.19013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8423736572266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15830, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09989204704761505, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.013837880641222, + "grad_norm": 0.13587051630020142, + "kept_token_count": 736.546875, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9925589919090271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441027276217938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.44549255371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.151708984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15840, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.10005679130554199, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.015164996311068536, + "grad_norm": 0.10476623475551605, + "kept_token_count": 736.3375, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9930261552333832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069738409481942655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7504486083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.37694091796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15850, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.09969286769628524, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.017892222665250302, + "grad_norm": 0.11656967550516129, + "kept_token_count": 736.75, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.99276202917099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007237966265529394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6337188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2679718017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15860, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.10058024600148201, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.014479556865990163, + "grad_norm": 0.09812798351049423, + "kept_token_count": 735.8796875, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9929172158241272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070827738381922245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.14650268554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.4576416015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15870, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.10002275854349137, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.014863465540111065, + "grad_norm": 0.1121593713760376, + "kept_token_count": 736.4421875, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9928900539875031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007109938655048609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.08861083984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.36346740722655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15880, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09957489296793938, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.016713680420070886, + "grad_norm": 0.13097886741161346, + "kept_token_count": 736.4546875, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9930927276611328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006907275505363941, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.70274658203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.62852172851564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15890, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.09940696209669113, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.013654362130910157, + "grad_norm": 0.11283255368471146, + "kept_token_count": 736.7890625, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929100632667541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007089937385171652, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2598602294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.95531005859374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15900, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.10020901411771774, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.013814164325594902, + "grad_norm": 0.13388974964618683, + "kept_token_count": 736.559375, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9929688572883606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031133584678173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2713897705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0008575439453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15910, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.10006786212325096, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.014066941291093826, + "grad_norm": 0.1309286653995514, + "kept_token_count": 736.6453125, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927035570144653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007296426501125097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.43955993652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1978790283203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15920, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.10012531131505967, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.014883750677108764, + "grad_norm": 0.13259385526180267, + "kept_token_count": 736.553125, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9920935750007629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007906428910791874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.52662353515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6560882568359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15930, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.10062131732702255, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.013207164080813526, + "grad_norm": 0.11756429821252823, + "kept_token_count": 736.1421875, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9933087527751923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006691252812743187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1745269775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.90929260253904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15940, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10004540979862213, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.0158251472748816, + "grad_norm": 0.12411811947822571, + "kept_token_count": 736.5515625, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9931476891040802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006852312013506889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1901885986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7726531982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15950, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.10006730183959008, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.013540377467870712, + "grad_norm": 0.08208487182855606, + "kept_token_count": 736.578125, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.992682695388794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007317309174686671, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4236785888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.07390747070315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15960, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.09970132261514664, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.014220945071429014, + "grad_norm": 0.10851536691188812, + "kept_token_count": 736.925, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9932338476181031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067661396227777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.09320373535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6428985595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15970, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.10081496387720108, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.015463293064385653, + "grad_norm": 0.1387454867362976, + "kept_token_count": 735.771875, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9928799450397492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007120069582015276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.36732788085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.79588317871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15980, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10049500092864036, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.014310123957693576, + "grad_norm": 0.1270752102136612, + "kept_token_count": 736.171875, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928948998451232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007105100061744451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.99266052246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.54771728515624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 15990, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.09977664276957512, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.01316203474998474, + "grad_norm": 0.0996558666229248, + "kept_token_count": 736.90625, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9931378901004791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006862108781933785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6394836425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5473175048828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16000, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.09925502240657806, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.012642095051705837, + "grad_norm": 0.11576583236455917, + "kept_token_count": 736.840625, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9926180422306061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007381959538906813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0538330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0797149658203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16010, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.10024391636252403, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.015240978915244341, + "grad_norm": 0.13339492678642273, + "kept_token_count": 736.0609375, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9925474524497986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007452544663101435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3448120117188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.76613159179686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16020, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.0998503640294075, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.012555933324620128, + "grad_norm": 0.0993397906422615, + "kept_token_count": 736.7875, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9927949666976928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007205029204487801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.08689880371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6429809570312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16030, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.10040677338838577, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.013812230713665486, + "grad_norm": 0.08322723954916, + "kept_token_count": 736.175, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9930580139160157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069419764913618565, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.37505798339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.18433532714846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16040, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.09970879554748535, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.014668758679181337, + "grad_norm": 0.11492004990577698, + "kept_token_count": 736.5546875, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9927766382694244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007223363034427166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2746276855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6277160644531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.09961977899074555, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.015714619867503644, + "grad_norm": 0.18502527475357056, + "kept_token_count": 736.4734375, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9930749595165252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069250441156327724, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3662933349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8378112792969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16060, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.10034130364656449, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.013476253813132643, + "grad_norm": 0.11281274259090424, + "kept_token_count": 736.30625, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9929340958595276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007065901719033718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8284027099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6294769287109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.10008233860135078, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.014928147383034229, + "grad_norm": 0.10197733342647552, + "kept_token_count": 736.178125, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9930024743080139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997524667531252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.82562255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9923431396484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16080, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.10014693811535835, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.01738663287833333, + "grad_norm": 0.09017174690961838, + "kept_token_count": 736.05, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9930618703365326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006938132084906101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6031890869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.88933715820315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16090, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.09968416169285774, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.011509142303839325, + "grad_norm": 0.10729168355464935, + "kept_token_count": 736.6875, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9926222443580628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007377751637250185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.52830505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9290710449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16100, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10010241940617562, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.013516524340957403, + "grad_norm": 0.08822818845510483, + "kept_token_count": 736.1703125, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9929581940174103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00704180896282196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.05048217773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.87709350585936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16110, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.10017983913421631, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.013428402040153742, + "grad_norm": 0.09620470553636551, + "kept_token_count": 736.3359375, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.992556232213974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007443763222545386, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0971160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3327941894531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16120, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09957675188779831, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.013830535486340523, + "grad_norm": 0.1328543722629547, + "kept_token_count": 736.7, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9926279067993165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007372088357806205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.70094299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4230194091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16130, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.0991691917181015, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.014518306590616703, + "grad_norm": 0.09620121121406555, + "kept_token_count": 737.05625, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927357971668244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007264191005378962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7862945556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7148712158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16140, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.1003459133207798, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.014006422273814679, + "grad_norm": 0.11999896168708801, + "kept_token_count": 735.7, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9930600523948669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069399290718138214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4349334716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7604064941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16150, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.10026895552873612, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.012968621030449868, + "grad_norm": 0.10682662576436996, + "kept_token_count": 735.9328125, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9932318270206452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006768158264458179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.94529418945314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16160, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.09980272203683853, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.013888995442539454, + "grad_norm": 0.11370663344860077, + "kept_token_count": 736.7703125, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9928583621978759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007141637243330478, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.05226135253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.62424621582034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.10091413483023644, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.015022735763341188, + "grad_norm": 0.13921387493610382, + "kept_token_count": 735.7578125, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926416337490082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00735837435349822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.37843017578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.13434143066405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16180, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.1003916546702385, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.014793759398162364, + "grad_norm": 0.15309825539588928, + "kept_token_count": 736.1265625, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9933411061763764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006658892799168825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4434356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.35162353515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16190, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.0996178299188614, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.016507836990058422, + "grad_norm": 0.0876072496175766, + "kept_token_count": 736.61875, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9926131010055542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007386905699968338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.62855834960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1609649658203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16200, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.1004503272473812, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.01526055745780468, + "grad_norm": 0.11098781228065491, + "kept_token_count": 736.0765625, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9927036106586457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007296386454254389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.93092041015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.18544311523436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16210, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09926654547452926, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.015172299928963184, + "grad_norm": 0.1117846816778183, + "kept_token_count": 736.9671875, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9932063162326813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067936996929347515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.15294494628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9267883300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16220, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.09947323054075241, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.012506422493606806, + "grad_norm": 0.08870085328817368, + "kept_token_count": 736.815625, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9931075572967529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006892456300556659, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.08665466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2133819580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.09996630176901818, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.01126588243059814, + "grad_norm": 0.08906979858875275, + "kept_token_count": 736.7203125, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9932872653007507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006712743267416954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.7873229980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.15997009277345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16240, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.10017655640840531, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.015346036991104484, + "grad_norm": 0.1361686885356903, + "kept_token_count": 736.4296875, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9934568703174591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006543135084211826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4342956542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.87767333984374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16250, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.100401172041893, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.014406490186229349, + "grad_norm": 0.10479813814163208, + "kept_token_count": 735.9546875, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929307758808136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007069207355380058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6443603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.67311096191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16260, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.09995497912168502, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.015713394340127706, + "grad_norm": 0.13204748928546906, + "kept_token_count": 736.6828125, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9931328177452088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006867172382771969, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.3177825927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8496795654297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16270, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.10019123256206512, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.014570446778088808, + "grad_norm": 0.09432759881019592, + "kept_token_count": 736.4234375, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927531123161316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007246885821223259, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.34413757324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9912536621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16280, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.1001644566655159, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.018512267526239155, + "grad_norm": 0.10704295337200165, + "kept_token_count": 736.265625, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9928246438503265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007175363786518574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9311157226563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9699340820313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.10033333003520965, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.015365776233375072, + "grad_norm": 0.09980010241270065, + "kept_token_count": 736.4859375, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9930463612079621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953649409115315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.75709228515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7934539794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16300, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.09951798841357232, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.013584297522902489, + "grad_norm": 0.1255643516778946, + "kept_token_count": 736.6296875, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930418372154236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069581735879182816, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.71482238769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.40928955078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16310, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.1000880055129528, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.014286504499614239, + "grad_norm": 0.10852539539337158, + "kept_token_count": 736.515625, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9929156541824341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007084347773343325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9253784179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.86476135253906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16320, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.10078815296292305, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.012934136390686034, + "grad_norm": 0.08050690591335297, + "kept_token_count": 735.865625, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9932446599006652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006755359563976526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0968505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.22730407714846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16330, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.09980016723275184, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.015578866750001908, + "grad_norm": 0.10198771953582764, + "kept_token_count": 736.478125, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9931145489215851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006885437574237585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2319671630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.74998474121094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16340, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09957820400595666, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.013359876908361911, + "grad_norm": 0.12485003471374512, + "kept_token_count": 736.7546875, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.993100768327713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006899225525557995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.14736022949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.98297424316405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16350, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09974984899163246, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.015109606366604567, + "grad_norm": 0.14515873789787292, + "kept_token_count": 736.553125, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9929569363594055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007043072115629911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.43016662597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6649444580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16360, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.10038366690278053, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.012801775243133307, + "grad_norm": 0.08667909353971481, + "kept_token_count": 736.1265625, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9932470679283142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006752929650247097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.71641235351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4474853515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16370, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.1005946785211563, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.013689077086746692, + "grad_norm": 0.11286791414022446, + "kept_token_count": 735.9, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929011583328247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007098838873207569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5966491699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.09873046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16380, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10017098635435104, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.01596219101920724, + "grad_norm": 0.10733793675899506, + "kept_token_count": 736.2609375, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925751566886902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00742483027279377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9460083007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.83179626464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16390, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09962536841630935, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.013108846079558135, + "grad_norm": 0.10478710383176804, + "kept_token_count": 736.64375, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.993261706829071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006738297641277313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4805389404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.47496643066404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16400, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10064435377717018, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.013621255289763212, + "grad_norm": 0.09131522476673126, + "kept_token_count": 735.95, + "learning_rate": 1.39113214589617e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9932477712631226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006752241589128971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.95562438964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3005767822266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16410, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.0995807982981205, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.013439717888832092, + "grad_norm": 0.09348353743553162, + "kept_token_count": 736.796875, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930435240268707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006956461071968079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.9108032226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0194610595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.09955641925334931, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.01418452663347125, + "grad_norm": 0.1072612851858139, + "kept_token_count": 737.0125, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.993198812007904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068011898547410965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7106414794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.75917358398436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16430, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.09979564547538758, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.012668930552899837, + "grad_norm": 0.08336471766233444, + "kept_token_count": 736.740625, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9934915125370025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006508494634181261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7874725341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.88523559570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16440, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.10024914667010307, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.01491080243140459, + "grad_norm": 0.10234139859676361, + "kept_token_count": 736.1734375, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927686750888824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007231323421001435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.8983551025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2796630859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16450, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.10042937770485878, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.011888586543500423, + "grad_norm": 0.09643291682004929, + "kept_token_count": 736.340625, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9931252837181092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006874699238687753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.08554992675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7678649902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16460, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.10024956613779068, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.01294441856443882, + "grad_norm": 0.08340434730052948, + "kept_token_count": 736.4703125, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929889738559723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007011035457253456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 460.1929901123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.05885314941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16470, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.10017264932394028, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.016187033057212828, + "grad_norm": 0.11307423561811447, + "kept_token_count": 736.26875, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9928686320781708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007131368294358254, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.47288818359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8973846435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16480, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.10009837076067925, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.01480792872607708, + "grad_norm": 0.13051837682724, + "kept_token_count": 736.4890625, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9930998146533966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006900183763355017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.96763000488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6599822998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16490, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10000187158584595, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.011166244093328715, + "grad_norm": 0.1244664266705513, + "kept_token_count": 736.5890625, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9930810630321503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006918933521956205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4678558349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6079895019531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16500, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09987103939056396, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.014233501721173525, + "grad_norm": 0.1402602344751358, + "kept_token_count": 736.71875, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927183270454407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007281663455069065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.3997863769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4700073242187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16510, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.09959492087364197, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.012880389019846917, + "grad_norm": 0.11965540796518326, + "kept_token_count": 736.6703125, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9933396995067596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006660290993750095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.6278137207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0727813720703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16520, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10007069930434227, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.014117983728647232, + "grad_norm": 0.14583346247673035, + "kept_token_count": 736.446875, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9928374469280243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007162570115178823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.06053161621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.35215454101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16530, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.10084781795740128, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.012241558311507106, + "grad_norm": 0.08610653877258301, + "kept_token_count": 735.7765625, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9926328420639038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007367163896560669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.28485412597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.441943359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.09964660555124283, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.013863332336768508, + "grad_norm": 0.09154773503541946, + "kept_token_count": 736.9109375, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9931444227695465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006855576764792204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3401611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0620971679688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16550, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.10023507326841355, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.013495046831667423, + "grad_norm": 0.08794035762548447, + "kept_token_count": 736.3046875, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930952191352844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006904787849634886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0640808105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6479736328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.10005706399679185, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.014025308564305306, + "grad_norm": 0.08457869291305542, + "kept_token_count": 736.5671875, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926377713680268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007362238597124815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.2935791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.71941528320315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16570, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.10016575381159783, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.014967559836804867, + "grad_norm": 0.1298464983701706, + "kept_token_count": 736.9875, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9929134547710419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007086563855409622, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.97320556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2874389648438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16580, + "valid_token_count": 819.025 + }, + { + "actual_token_mask_ratio": 0.10060193985700608, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.014589531347155571, + "grad_norm": 0.10249611735343933, + "kept_token_count": 735.7296875, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9930802762508393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006919715460389853, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.24103393554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.87611083984376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16590, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.0995853878557682, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.016458342922851442, + "grad_norm": 0.08863578736782074, + "kept_token_count": 737.0890625, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9926254987716675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007374495454132557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.36322021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5728240966797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16600, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.1006332129240036, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.013643622072413564, + "grad_norm": 0.13590578734874725, + "kept_token_count": 735.9734375, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929317951202392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00706821633502841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.70116271972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.78160400390624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16610, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.10028787776827812, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.014307608548551797, + "grad_norm": 0.1011577919125557, + "kept_token_count": 736.065625, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992779278755188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00722070774063468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.01246337890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.08001403808595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16620, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.0998370498418808, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.013415630999952554, + "grad_norm": 0.08927800506353378, + "kept_token_count": 736.7359375, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9933571577072143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642849463969469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0360931396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2305206298828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16630, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.09966210648417473, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.01650160225108266, + "grad_norm": 0.14761757850646973, + "kept_token_count": 736.9578125, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9926727592945099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007327241729944945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4816162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4817291259766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16640, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.09995129108428955, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.013647852838039399, + "grad_norm": 0.15034319460391998, + "kept_token_count": 736.4109375, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929000496864319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007099945563822985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4069091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.49609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16650, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.10030915960669518, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.014551069494336844, + "grad_norm": 0.13423095643520355, + "kept_token_count": 736.3078125, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9930678248405457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006932174414396286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.7318969726563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0255584716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16660, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.09958468824625015, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.01601812904700637, + "grad_norm": 0.09652985632419586, + "kept_token_count": 736.8375, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9927639424800873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007236065156757831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9749694824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.24585876464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10005996897816657, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.014445091225206852, + "grad_norm": 0.11848177760839462, + "kept_token_count": 736.640625, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.993153864145279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068461280316114426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5364196777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0922607421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16680, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.09950853437185288, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.01737365871667862, + "grad_norm": 0.10997960716485977, + "kept_token_count": 736.971875, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9929660439491272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007033962290734053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4553955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.33681640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16690, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.09997234120965004, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.013452100101858377, + "grad_norm": 0.09706516563892365, + "kept_token_count": 736.6734375, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930582821369172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941736955195665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.28765869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.92307739257814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16700, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.10043370947241784, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.011982588190585374, + "grad_norm": 0.10638856142759323, + "kept_token_count": 736.203125, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932893216609955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710668373852968, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.90288696289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3635772705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16710, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.09937377423048019, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.012885961635038256, + "grad_norm": 0.10538549721240997, + "kept_token_count": 736.9921875, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9931267261505127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006873266957700253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3354797363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8645294189453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16720, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.10024078711867332, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.013726085936650635, + "grad_norm": 0.14851124584674835, + "kept_token_count": 736.2296875, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9927947044372558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007205292955040932, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0234680175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7245758056641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16730, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09987341165542603, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.012834066245704889, + "grad_norm": 0.10115128010511398, + "kept_token_count": 736.8484375, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9934327244758606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006567256711423397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2438079833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9936981201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16740, + "valid_token_count": 818.609375 + }, + { + "actual_token_mask_ratio": 0.10016550421714783, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.015407089702785015, + "grad_norm": 0.10325074940919876, + "kept_token_count": 736.4125, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9932071149349213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067929021082818505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0431182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7052947998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16750, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.09999868795275688, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.013260337244719266, + "grad_norm": 0.13396823406219482, + "kept_token_count": 736.821875, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931023716926575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068976316601037976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1386657714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.92686157226564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16760, + "valid_token_count": 818.6890625 + }, + { + "actual_token_mask_ratio": 0.09939064607024192, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.014295822568237781, + "grad_norm": 0.10478970408439636, + "kept_token_count": 737.38125, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9933721482753753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006627853028476238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.36114807128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4680419921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16770, + "valid_token_count": 818.7546875 + }, + { + "actual_token_mask_ratio": 0.10109184831380844, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.013919569831341506, + "grad_norm": 0.11630583554506302, + "kept_token_count": 735.71875, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9932858109474182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006714181415736676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5314880371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8967742919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16780, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.09968278929591179, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.01313155135139823, + "grad_norm": 0.07535222917795181, + "kept_token_count": 736.6, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9933796405792237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066203616559505464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0187042236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.54149169921874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16790, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09986005648970604, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.014726462308317423, + "grad_norm": 0.13856402039527893, + "kept_token_count": 736.6921875, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9931819021701813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006818096153438091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3850067138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5440338134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16800, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.09914014413952828, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.014288251753896475, + "grad_norm": 0.14330081641674042, + "kept_token_count": 737.11875, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9936733305454254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006326669733971358, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.50242004394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8607177734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09904530346393585, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.014475512225180865, + "grad_norm": 0.10333476960659027, + "kept_token_count": 737.14375, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9933712184429169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066287854686379434, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.68695373535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.0534240722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16820, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.1000610575079918, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.015655255410820246, + "grad_norm": 0.14387273788452148, + "kept_token_count": 736.3375, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9929867506027221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007013244181871414, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.46190490722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5649108886719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16830, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10029076486825943, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.012736630719155073, + "grad_norm": 0.11383214592933655, + "kept_token_count": 736.5984375, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932157337665558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006784254778176546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9561309814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5887451171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16840, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.10048217624425888, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.013433645665645599, + "grad_norm": 0.09954670071601868, + "kept_token_count": 736.1625, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930119931697845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006988007202744484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8137237548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9664337158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.10035960376262665, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.013432934274896979, + "grad_norm": 0.1348685622215271, + "kept_token_count": 736.10625, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9927124917507172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007287501730024815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.92430725097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.96128845214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16860, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.09922178611159324, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.013781478675082327, + "grad_norm": 0.12658484280109406, + "kept_token_count": 737.1375, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9930476546287537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006952350214123726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5024383544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.43938903808595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16870, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10012225806713104, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.012939503975212574, + "grad_norm": 0.08291333168745041, + "kept_token_count": 736.559375, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9925420641899109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074579334817826744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9632049560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.62103271484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16880, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.10023137256503105, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.012502446863800287, + "grad_norm": 0.08376391232013702, + "kept_token_count": 736.4453125, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9936566233634949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00634337067604065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.0983489990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.64711303710936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16890, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.10013722702860832, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.014737639110535383, + "grad_norm": 0.10621152818202972, + "kept_token_count": 736.1, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928779125213623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007122074719518423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.25987548828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.45314636230466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16900, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.10050011277198792, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.012320697028189897, + "grad_norm": 0.09226366132497787, + "kept_token_count": 736.45625, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9937875032424927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062124903313815595, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.12567749023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4966583251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16910, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.10001181438565254, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.013390012923628092, + "grad_norm": 0.1154206395149231, + "kept_token_count": 736.4796875, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9931831359863281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006816861499100923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8670135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.45684814453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16920, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.09947514981031418, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.013046490121632814, + "grad_norm": 0.10169245302677155, + "kept_token_count": 736.9875, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9927227199077606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072772840037941934, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8190399169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.75672912597656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16930, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.09932201951742173, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.012228564638644458, + "grad_norm": 0.11560641974210739, + "kept_token_count": 737.0765625, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930256724357605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006974325142800808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.34681091308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.64581604003905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16940, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09985724613070487, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.012862567510455846, + "grad_norm": 0.11536905169487, + "kept_token_count": 736.609375, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929962635040284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070037486962974075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0369506835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2094268798828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16950, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.09962795227766037, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.011369848158210516, + "grad_norm": 0.09886632114648819, + "kept_token_count": 736.746875, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9933256268501282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00667437743395567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2246368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2926971435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16960, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.10020269453525543, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.012952040415257215, + "grad_norm": 0.06579358130693436, + "kept_token_count": 736.475, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9925466358661652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007453367207199335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3061248779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8321838378906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16970, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.09999872148036956, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.014128803368657827, + "grad_norm": 0.11379794776439667, + "kept_token_count": 736.284375, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9929599165916443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007040084060281515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9473114013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.0762603759766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16980, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.09974392354488373, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.015003155684098601, + "grad_norm": 0.10232095420360565, + "kept_token_count": 736.7203125, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9931593418121338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006840668618679047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6645111083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8059936523438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 16990, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09985696524381638, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.014505642745643855, + "grad_norm": 0.13045737147331238, + "kept_token_count": 736.43125, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9933370888233185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006662899814546108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3606903076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.47589721679685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17000, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.09953028559684754, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.015120532270520926, + "grad_norm": 0.10823032259941101, + "kept_token_count": 736.6859375, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9925490379333496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007450963463634252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1385498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.62083740234374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17010, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.10018131583929062, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.01346534164622426, + "grad_norm": 0.1136937141418457, + "kept_token_count": 736.35625, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9927011430263519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007298852410167455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0814727783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9170654296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17020, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09972111284732818, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.012808656971901655, + "grad_norm": 0.09532935917377472, + "kept_token_count": 736.5703125, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9933117926120758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00668821120634675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1509155273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.02178955078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17030, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09951354935765266, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.014266967866569757, + "grad_norm": 0.1047879308462143, + "kept_token_count": 736.7328125, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927789747714997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007221037894487381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.80808410644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0831024169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17040, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.09934827759861946, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.013190783560276031, + "grad_norm": 0.10947926342487335, + "kept_token_count": 736.9109375, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930078089237213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006992197595536709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9544250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0319488525391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17050, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09986501187086105, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014481118693947793, + "grad_norm": 0.09871447086334229, + "kept_token_count": 736.3875, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9928340375423431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007165955286473036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.7065856933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.86571044921874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17060, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09990492537617683, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.012047814764082431, + "grad_norm": 0.10904175043106079, + "kept_token_count": 736.4703125, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9925869524478912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007413060311228037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.5280059814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.1115447998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17070, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.09955647513270378, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.014397948514670134, + "grad_norm": 0.12087511271238327, + "kept_token_count": 736.8828125, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9933659851551055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006634014006704092, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1549041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.62086486816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09927807599306107, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.015335210599005222, + "grad_norm": 0.1078166589140892, + "kept_token_count": 736.9046875, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9930587112903595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941268779337406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.30376892089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7593719482422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17090, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.09985716491937638, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.012693600449711084, + "grad_norm": 0.11357540637254715, + "kept_token_count": 736.3265625, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9930888950824738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069111064076423645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0702880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4293182373047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.09958850592374802, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.01580506069585681, + "grad_norm": 0.10738811641931534, + "kept_token_count": 736.6984375, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9930933237075805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906675081700087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.3733184814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8757080078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17110, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.09996213987469674, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.012030203128233552, + "grad_norm": 0.09819570928812027, + "kept_token_count": 736.6109375, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9932243227958679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006775669939815998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.8363311767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.20648803710935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17120, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.10020276829600334, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.01440702537074685, + "grad_norm": 0.08158477395772934, + "kept_token_count": 736.3875, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.993386173248291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066138354130089285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5480194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.4194396972656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17130, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.10025211051106453, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015006990544497967, + "grad_norm": 0.09726830571889877, + "kept_token_count": 735.8703125, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9933968007564544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066032025031745436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4047088623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.19044189453126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17140, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.0994440883398056, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.0130663619376719, + "grad_norm": 0.09861503541469574, + "kept_token_count": 736.578125, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9930640995502472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006935890857130289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.47313537597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2817352294922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17150, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.09975834712386131, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.013293268950656056, + "grad_norm": 0.08161354809999466, + "kept_token_count": 736.753125, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9933867990970612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613209471106529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.05816955566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4988067626953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17160, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.10010493099689484, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.013142004795372485, + "grad_norm": 0.12419341504573822, + "kept_token_count": 736.2171875, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9933347642421723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006665232591331005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5413452148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9582824707031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17170, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.1004337377846241, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.015238136984407902, + "grad_norm": 0.15492664277553558, + "kept_token_count": 736.1625, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9933580696582794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006641934625804424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9397918701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.03965454101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17180, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.09989324510097504, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013117681536823512, + "grad_norm": 0.11053895205259323, + "kept_token_count": 736.4671875, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9932944536209106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067055530846118925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2989501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.12562255859376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17190, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.0996399812400341, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.013249224703758954, + "grad_norm": 0.09206479042768478, + "kept_token_count": 736.5125, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9928507447242737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071492623537778854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.21747131347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3846893310547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17200, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.1004722073674202, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.013088608533143998, + "grad_norm": 0.12259253859519958, + "kept_token_count": 736.20625, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929631054401398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070368914864957334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4303375244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5364501953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17210, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.09940262511372566, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.015471069514751435, + "grad_norm": 0.13281157612800598, + "kept_token_count": 736.75, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9931553661823272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068446222692728044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.65362548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8987762451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17220, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.09965045601129532, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.014616721495985984, + "grad_norm": 0.10691250115633011, + "kept_token_count": 736.6453125, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9930999338626861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069000743329525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9729309082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.86464538574216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17230, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.0998819425702095, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.012910893047228455, + "grad_norm": 0.10493535548448563, + "kept_token_count": 736.95, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9933491766452789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006650841049849987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.195703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6785552978516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17240, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.10016103908419609, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.01410915507003665, + "grad_norm": 0.10947337001562119, + "kept_token_count": 736.3390625, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9931930422782898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006806960981339216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9555358886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.09938354492186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17250, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.09966627955436706, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.015491543058305978, + "grad_norm": 0.1147979348897934, + "kept_token_count": 736.5, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9928712606430053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007128738611936569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.63560791015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.78846435546876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17260, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.10083028227090836, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.012446541339159012, + "grad_norm": 0.11948901414871216, + "kept_token_count": 735.9953125, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.016, + "mgd_cos_sim": 0.992865926027298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071340627036988735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.79326782226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.55419921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17270, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.09920761734247208, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.013444780092686414, + "grad_norm": 0.08246833086013794, + "kept_token_count": 736.9578125, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9938073813915252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00619261572137475, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8870910644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3204071044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10018425807356834, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.01294455686584115, + "grad_norm": 0.0808524638414383, + "kept_token_count": 735.796875, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9927808284759522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00721917562186718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.42454528808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2200897216797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17290, + "valid_token_count": 817.721875 + }, + { + "actual_token_mask_ratio": 0.10006433054804802, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.013312329817563296, + "grad_norm": 0.0864953100681305, + "kept_token_count": 736.5578125, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931249797344208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006875027995556593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.63673095703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1081481933594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17300, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.09977470487356185, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.01507695559412241, + "grad_norm": 0.10632438957691193, + "kept_token_count": 736.7640625, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926986694335938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007301343884319067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6803283691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.75115356445315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17310, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.09936264380812646, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.014218493457883596, + "grad_norm": 0.10515514016151428, + "kept_token_count": 736.665625, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9930538833141327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006946103740483522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2070251464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.831689453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17320, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.10028575882315635, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.0154762402176857, + "grad_norm": 0.10157570242881775, + "kept_token_count": 736.3078125, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927960634231567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007203916646540165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1524688720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2776245117187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17330, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.1002460964024067, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.011790894577279687, + "grad_norm": 0.09604893624782562, + "kept_token_count": 736.1765625, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9929280817508698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070719138719141485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1632110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.33094787597656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17340, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09997502416372299, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.013264605402946472, + "grad_norm": 0.0985485389828682, + "kept_token_count": 736.41875, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931770861148834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006822919752448797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1173034667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1283660888672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09940917938947677, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.011992862308397889, + "grad_norm": 0.1193731427192688, + "kept_token_count": 737.0609375, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9930948555469513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006905141938477755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.53460998535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.81191101074216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17360, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.10021641030907631, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.014159627258777618, + "grad_norm": 0.08940080553293228, + "kept_token_count": 736.4171875, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.993133795261383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006866205111145973, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1048187255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0342712402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17370, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.10059363842010498, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.012461352813988923, + "grad_norm": 0.10933920741081238, + "kept_token_count": 736.1703125, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9931099474430084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006890057865530252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1578430175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6285797119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17380, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.10031700506806374, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.013791390508413316, + "grad_norm": 0.1028774082660675, + "kept_token_count": 735.9, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9930173039436341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006982701830565929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.307763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.97919006347655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17390, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.09934097602963447, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.015885370131582022, + "grad_norm": 0.11893096566200256, + "kept_token_count": 736.65, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9933539211750031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006646076869219542, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.14873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5113220214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17400, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.0997856505215168, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.0131698505487293, + "grad_norm": 0.08614768832921982, + "kept_token_count": 736.59375, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929774165153503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007022577524185181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.00722351074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1339111328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.09990685284137726, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.013095748797059058, + "grad_norm": 0.11050354689359665, + "kept_token_count": 736.2671875, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926084578037262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007391539961099624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.60545349121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.62928771972656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17420, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.10024054050445556, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.013129595248028637, + "grad_norm": 0.12078911811113358, + "kept_token_count": 736.4765625, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930449843406677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006955020409077406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.88393249511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1355499267578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17430, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.10038017705082894, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.011218462232500314, + "grad_norm": 0.09805957227945328, + "kept_token_count": 736.2546875, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9930688381195069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006931169331073761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.18467102050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8414367675781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17440, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.10124389454722404, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.011399815091863275, + "grad_norm": 0.11652399599552155, + "kept_token_count": 735.3234375, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9936352133750915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006364788580685854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.09813232421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3354858398437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17450, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09980710297822952, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.015236476343125105, + "grad_norm": 0.14173807203769684, + "kept_token_count": 736.4296875, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9930859267711639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006914067827165127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2012451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1279541015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17460, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.09985502585768699, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.013953064475208521, + "grad_norm": 0.10836324095726013, + "kept_token_count": 736.725, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9924022257328033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007597774174064398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.26015014648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3132019042969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17470, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.10043558701872826, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.013989395089447498, + "grad_norm": 0.10900925099849701, + "kept_token_count": 735.9296875, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9932294249534607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006770573277026415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1302459716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6434692382812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17480, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.0994475394487381, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.014683054946362972, + "grad_norm": 0.10226423293352127, + "kept_token_count": 736.8125, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9931829452514649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068170599639415744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.24720458984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.1305297851562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17490, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.0999602921307087, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.012266889307647943, + "grad_norm": 0.09079328924417496, + "kept_token_count": 736.371875, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.993544089794159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006455917004495859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.87961730957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9848602294922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17500, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10013115406036377, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.012014711368829012, + "grad_norm": 0.11124265938997269, + "kept_token_count": 735.9234375, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9936218380928039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006378173734992743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.46007080078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2571716308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17510, + "valid_token_count": 817.8109375 + }, + { + "actual_token_mask_ratio": 0.10056174471974373, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.015984599571675064, + "grad_norm": 0.11602329462766647, + "kept_token_count": 736.3140625, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925026893615723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007497310265898704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7665161132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.25562744140626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17520, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.0989920161664486, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.01581459017470479, + "grad_norm": 0.0949382483959198, + "kept_token_count": 737.153125, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.993258786201477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006741211283951998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5794677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.66965942382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17530, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.09997022822499275, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.01521214134991169, + "grad_norm": 0.0869135856628418, + "kept_token_count": 736.46875, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9930979073047638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069020787253975865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7933837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.57481079101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.09866650998592377, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.013457100139930844, + "grad_norm": 0.13612942397594452, + "kept_token_count": 737.1265625, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9933127880096435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006687215156853199, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6933227539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5318084716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17550, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.09999308064579963, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.01358315274119377, + "grad_norm": 0.0806027352809906, + "kept_token_count": 736.2546875, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.992877197265625, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007122780941426754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.08266296386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.513134765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17560, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.09993405118584633, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.014568210067227482, + "grad_norm": 0.08612979203462601, + "kept_token_count": 736.7828125, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9931924879550934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006807514652609825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.61378479003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6128631591797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17570, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.0993984304368496, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.014584099501371383, + "grad_norm": 0.10129835456609726, + "kept_token_count": 736.6109375, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9931952118873596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006804786529392004, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.38812255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5513427734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17580, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.0992576651275158, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.014665672648698091, + "grad_norm": 0.10950540751218796, + "kept_token_count": 736.9921875, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9931171298027038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006882880255579949, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.06769409179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.59975280761716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17590, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.10008966401219369, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.013389916718006134, + "grad_norm": 0.10893315076828003, + "kept_token_count": 736.2390625, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9932082772254944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791722681373358, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1093353271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2913452148438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17600, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.09965246692299842, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.012670076172798871, + "grad_norm": 0.09174393862485886, + "kept_token_count": 736.8515625, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9933485805988311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006651419401168823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8564697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.78497314453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17610, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.0999000146985054, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.011538376472890377, + "grad_norm": 0.08372634649276733, + "kept_token_count": 736.6484375, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9930770695209503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922922469675541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8317077636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6388305664062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.09993020370602608, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.01629849374294281, + "grad_norm": 0.1066211611032486, + "kept_token_count": 736.7484375, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9925540268421174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007445962261408567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.07997741699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8345611572266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17630, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.10008688569068909, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.014950533024966716, + "grad_norm": 0.13307638466358185, + "kept_token_count": 736.4953125, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9930947721004486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006905227154493332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1925079345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6427795410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17640, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10051709935069084, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.014094282081350685, + "grad_norm": 0.077926404774189, + "kept_token_count": 735.9859375, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9932755649089813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006724444683641195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.40184631347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.87793884277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17650, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09996051192283631, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.014041109662503003, + "grad_norm": 0.07683275640010834, + "kept_token_count": 736.8125, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9924647390842438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075352763757109645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0911529541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.305126953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17660, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.09979833588004113, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.011076834378764033, + "grad_norm": 0.10587113350629807, + "kept_token_count": 736.59375, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9929385423660279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007061449531465769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8734130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68388671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17670, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10033988282084465, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.01257416671141982, + "grad_norm": 0.10869430005550385, + "kept_token_count": 736.003125, + "learning_rate": 1.183443786387094e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9930715799331665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006928428821265697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.176220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.53525695800784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17680, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.10009485259652137, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.014772590529173613, + "grad_norm": 0.11488896608352661, + "kept_token_count": 736.5765625, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.993107920885086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068920918740332125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.4597625732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.94993286132814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17690, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.10007746368646622, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.012785254046320915, + "grad_norm": 0.10254558175802231, + "kept_token_count": 736.290625, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9931263208389283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006873689778149128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5561767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.6482208251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17700, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.09985171109437943, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.012444702628999948, + "grad_norm": 0.09593097865581512, + "kept_token_count": 736.3828125, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928122520446777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187731936573983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.75165405273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9902740478516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17710, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.1000747986137867, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.013586659356951714, + "grad_norm": 0.10285631567239761, + "kept_token_count": 736.59375, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9929641127586365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007035877555608749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.20542602539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.8595794677734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17720, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.10009942501783371, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.013095870381221175, + "grad_norm": 0.08616673946380615, + "kept_token_count": 736.509375, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9933146238327026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006685369648039341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.14732971191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.2735961914062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17730, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.09978781715035438, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.01231415504589677, + "grad_norm": 0.11172661185264587, + "kept_token_count": 736.6125, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9931656002998352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006834407430142164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.63177185058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9736572265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17740, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.09935162737965583, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.013883655611425638, + "grad_norm": 0.11987390369176865, + "kept_token_count": 737.225, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9933875262737274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612477172166109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.21458435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.1898651123047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17750, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.10037304237484931, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.011423878651112319, + "grad_norm": 0.08769549429416656, + "kept_token_count": 736.240625, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9930549800395966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006945012975484133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1001922607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3622161865234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17760, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.09949829056859016, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.012904720474034547, + "grad_norm": 0.12233622372150421, + "kept_token_count": 736.821875, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929923593997956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007007650472223759, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.51829223632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0579467773438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17770, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09981803894042969, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.01349599463865161, + "grad_norm": 0.1413174271583557, + "kept_token_count": 736.68125, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9931702196598053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006829775497317314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1245147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2178436279297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17780, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.0995771899819374, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.013175560720264911, + "grad_norm": 0.09049249440431595, + "kept_token_count": 736.6734375, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9933098137378693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006690183933824301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5377624511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3089599609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17790, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.10014860779047012, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.01315502505749464, + "grad_norm": 0.0855017676949501, + "kept_token_count": 736.475, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930485367774964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006951468903571368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1165496826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0460174560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17800, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.1004903994500637, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.014012798760086298, + "grad_norm": 0.10661077499389648, + "kept_token_count": 736.015625, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9933546006679534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006645398121327162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0542755126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9128387451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09955302774906158, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.015167967695742846, + "grad_norm": 0.11615259200334549, + "kept_token_count": 736.5484375, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9930010080337525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006998995877802372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.98109436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.24583129882814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17820, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.10008312612771988, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.013276585098356008, + "grad_norm": 0.15455208718776703, + "kept_token_count": 736.35625, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9932612419128418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006738759111613035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.188916015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.13645935058594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17830, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09965980350971222, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.014139877166599035, + "grad_norm": 0.09845022857189178, + "kept_token_count": 736.73125, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9924357950687408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007564203720539809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3003723144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 403.2559387207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17840, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.10030710697174072, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.014354062266647816, + "grad_norm": 0.09447299689054489, + "kept_token_count": 735.88125, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9932240068912506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006776006985455751, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.15758361816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6147064208984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17850, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.10065300688147545, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.013282254803925753, + "grad_norm": 0.1134011298418045, + "kept_token_count": 736.103125, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9934715628623962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006528440210968256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0949768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2559326171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17860, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.09939533397555352, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.014708478469401598, + "grad_norm": 0.11347953230142593, + "kept_token_count": 737.1390625, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9933068454265594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006693162396550179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.89662780761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.80128479003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17870, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.0989770993590355, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013238478638231754, + "grad_norm": 0.0965767428278923, + "kept_token_count": 737.1625, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931210935115814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006878900900483131, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.64039611816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.73320922851565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17880, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.09999627247452736, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013375673582777381, + "grad_norm": 0.10549220442771912, + "kept_token_count": 736.3359375, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9929209887981415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007079006731510162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.4488830566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4245880126953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17890, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.10052529126405715, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012546639842912554, + "grad_norm": 0.08737866580486298, + "kept_token_count": 736.2125, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9936148405075074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006385151855647564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.62734069824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.99872131347655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17900, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.10014790296554565, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.012116639781743288, + "grad_norm": 0.08100291341543198, + "kept_token_count": 736.45625, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9935059189796448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006494075804948807, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.52501220703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3777709960938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17910, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.09967676103115082, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.015505674853920936, + "grad_norm": 0.1339477151632309, + "kept_token_count": 736.2765625, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927292466163635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007270748168230057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.5291351318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4291473388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17920, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.09906972348690032, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.01227381662465632, + "grad_norm": 0.09042594581842422, + "kept_token_count": 737.1140625, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933456182479858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006654395163059235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5041137695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.73802490234374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17930, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.10051275566220283, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.013569796131923795, + "grad_norm": 0.10069812089204788, + "kept_token_count": 735.9203125, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926857054233551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007314296904951334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.89251708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9490112304687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17940, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.10012404322624206, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.012401176011189818, + "grad_norm": 0.10821659862995148, + "kept_token_count": 736.234375, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928493082523346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007150689791887999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.53115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9739196777344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17950, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.09972193017601967, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.012667831592261791, + "grad_norm": 0.10962299257516861, + "kept_token_count": 736.73125, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9926457405090332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007354265358299017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.56128845214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17960, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.09994105249643326, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.015146546950563788, + "grad_norm": 0.09102831035852432, + "kept_token_count": 736.4, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.992684292793274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00731568867340684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5643005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6361938476563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1000742070376873, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.011060650320723652, + "grad_norm": 0.11549042910337448, + "kept_token_count": 736.459375, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9934028267860413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006597181782126427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.14832763671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.91347351074216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17980, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10026426389813423, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.011626373697072267, + "grad_norm": 0.1140863299369812, + "kept_token_count": 736.3703125, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9935025691986084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006497430801391602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.75311279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.68699645996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 17990, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.0995686948299408, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.013350499421358108, + "grad_norm": 0.13198725879192352, + "kept_token_count": 737.1671875, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9931692063808442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006830792687833309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1298492431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.10545959472654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18000, + "valid_token_count": 818.6828125 + }, + { + "actual_token_mask_ratio": 0.09959135502576828, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.012423170870169998, + "grad_norm": 0.132890984416008, + "kept_token_count": 736.709375, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.993345957994461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006654038093984127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8726745605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0101837158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.09994839280843734, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.012564543914049865, + "grad_norm": 0.08857207745313644, + "kept_token_count": 736.54375, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9933211088180542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006678899936378002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0699127197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1524200439453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18020, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10036289393901825, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.01337764309719205, + "grad_norm": 0.12076986581087112, + "kept_token_count": 736.2421875, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992993438243866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007006541453301906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.17267456054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1396209716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18030, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.09970019459724426, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.013713822606950999, + "grad_norm": 0.09590215235948563, + "kept_token_count": 736.540625, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9936981081962586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006301884260028601, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.49097595214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.0698699951172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18040, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.10023332238197327, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.013149907439947128, + "grad_norm": 0.09525009989738464, + "kept_token_count": 736.2421875, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9937105238437652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006289469450712204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.596435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 404.14637145996096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18050, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10026228725910187, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.012642718758434058, + "grad_norm": 0.10943968594074249, + "kept_token_count": 736.4859375, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9930792808532715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006920722220093012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6983367919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.35069274902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18060, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.0994976855814457, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.013048162683844566, + "grad_norm": 0.12365934997797012, + "kept_token_count": 736.915625, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9933192014694214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006680785305798053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4848571777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6201110839844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18070, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.10060305222868919, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.013933420740067958, + "grad_norm": 0.10898595303297043, + "kept_token_count": 735.928125, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9929805099964142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007019480597227812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.73853454589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3019134521484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18080, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.09990176260471344, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.014398637600243092, + "grad_norm": 0.11198422312736511, + "kept_token_count": 736.5890625, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929615199565888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038490567356348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9801361083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.84534912109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18090, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.0993054486811161, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.012430826667696238, + "grad_norm": 0.08131974935531616, + "kept_token_count": 736.925, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.016, + "mgd_cos_sim": 0.992961448431015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038556784391403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0234649658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7610809326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18100, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.10045199319720269, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.013134218659251929, + "grad_norm": 0.1085074171423912, + "kept_token_count": 735.965625, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926979124546051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007302079349756241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.26092224121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.37537536621096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18110, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.09948476254940034, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.013042272068560123, + "grad_norm": 0.1115705594420433, + "kept_token_count": 736.9375, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929343402385712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007065651752054691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4502014160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3396240234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.10030637755990028, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.012446392560377717, + "grad_norm": 0.10727085173130035, + "kept_token_count": 736.3453125, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9926495552062988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00735044814646244, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.6347961425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1210083007812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18130, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.10032597333192825, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.015972848888486622, + "grad_norm": 0.09153908491134644, + "kept_token_count": 736.146875, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9936336159706116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366380397230387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4688690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9813629150391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18140, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.10050920471549034, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.01209419807419181, + "grad_norm": 0.0936974361538887, + "kept_token_count": 735.8875, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9937697887420655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006230202689766884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.26674194335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3029022216797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18150, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.10045327916741371, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.012461620662361384, + "grad_norm": 0.08550340682268143, + "kept_token_count": 735.8375, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9930101335048676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006989873200654984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6211791992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2497039794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18160, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.09970027878880501, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.01595399444922805, + "grad_norm": 0.0763663500547409, + "kept_token_count": 736.675, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9931357204914093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006864278856664896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.8546905517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4868896484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18170, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09916499629616737, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.013547925371676683, + "grad_norm": 0.08875704556703568, + "kept_token_count": 737.040625, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9928793549537659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007120638899505138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.97133178710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.55015869140624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18180, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.09993370324373245, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.013548631919547915, + "grad_norm": 0.1077365130186081, + "kept_token_count": 736.2125, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9935661435127259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433855090290308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.66851196289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2830108642578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18190, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.10058425292372704, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.011824136134237051, + "grad_norm": 0.08539934456348419, + "kept_token_count": 736.284375, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9932275772094726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006772424839437008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.35010986328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.52898864746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18200, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.10041404739022255, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.012172062229365111, + "grad_norm": 0.10387560725212097, + "kept_token_count": 736.1, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931119620800019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068880260922014715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4162963867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3596160888672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18210, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.09996177330613136, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.015802773647010326, + "grad_norm": 0.1193644106388092, + "kept_token_count": 736.628125, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9930746495723725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00692535825073719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.38022766113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.70249938964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18220, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.0997892178595066, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.012640818348154425, + "grad_norm": 0.10122794657945633, + "kept_token_count": 736.309375, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9931583940982819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006841605622321367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3927001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9518615722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18230, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.09982274621725082, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.012860694248229266, + "grad_norm": 0.10681100189685822, + "kept_token_count": 736.3125, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930834412574768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069165765307843685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5309661865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9231842041016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18240, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.09975187405943871, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.013060924038290978, + "grad_norm": 0.10445506870746613, + "kept_token_count": 736.403125, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.993492043018341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065079454332590105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.4945098876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2905029296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.09961504712700844, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.013317648228257895, + "grad_norm": 0.12495075166225433, + "kept_token_count": 736.703125, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9927277982234954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007272210344672203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7163391113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.68919372558594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18260, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.09956763386726379, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.014954173378646374, + "grad_norm": 0.09328167885541916, + "kept_token_count": 736.7875, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9926529347896575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007347067911177873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.36272888183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7872344970703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18270, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.10031827092170716, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.012745818868279458, + "grad_norm": 0.09066520631313324, + "kept_token_count": 736.228125, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932701647281647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006729843188077211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5516632080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.97638549804685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18280, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.10053887590765953, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.013488495955243707, + "grad_norm": 0.10787586122751236, + "kept_token_count": 735.728125, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9931626617908478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068373410031199455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1415313720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4779968261719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18290, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.10012741386890411, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.012585888523608446, + "grad_norm": 0.10446825623512268, + "kept_token_count": 736.240625, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9932086050510407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791400443762541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7434509277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.96192626953126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18300, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.10010407269001007, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.014629250019788742, + "grad_norm": 0.10014576464891434, + "kept_token_count": 736.3703125, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9930700182914733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006929981149733066, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.92479553222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5814544677734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.09969313815236092, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.012577394023537636, + "grad_norm": 0.08051873743534088, + "kept_token_count": 736.7140625, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9929259061813355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007074095215648413, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0897705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8609588623047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18320, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.0999542884528637, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.016177768353372812, + "grad_norm": 0.10530146211385727, + "kept_token_count": 736.3359375, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.992771053314209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007228958047926426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5814971923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.27344665527346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18330, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.09957715272903442, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.013297302508726716, + "grad_norm": 0.11819206178188324, + "kept_token_count": 736.9875, + "learning_rate": 1.07770404996693e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9926361620426178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007363842055201531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2100006103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.79392395019534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18340, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.10031072497367859, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.011536062601953745, + "grad_norm": 0.10228897631168365, + "kept_token_count": 736.1625, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9936390936374664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006360891368240118, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6059295654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.38287658691405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18350, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.1001021422445774, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.012752936268225313, + "grad_norm": 0.15548749268054962, + "kept_token_count": 736.4359375, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932247817516326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006775227468460798, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.99358825683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.90899963378905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18360, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10010217577219009, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.013049403578042984, + "grad_norm": 0.12698793411254883, + "kept_token_count": 736.2875, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9932956516742706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006704352051019669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8188751220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.82550048828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18370, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.09953632950782776, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.015241739805787802, + "grad_norm": 0.11180789768695831, + "kept_token_count": 736.9265625, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9934426844120026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006557317543774843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.27403869628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5377990722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18380, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.10017412900924683, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.01169424462132156, + "grad_norm": 0.09902840852737427, + "kept_token_count": 736.56875, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9933763086795807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066236846148967745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4806304931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.47939147949216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18390, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.10025549158453942, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.01445276802405715, + "grad_norm": 0.17664162814617157, + "kept_token_count": 736.3140625, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.99306218624115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006937821209430694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8765350341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.66319885253904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18400, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.09960246682167054, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.014710263907909393, + "grad_norm": 0.11419212073087692, + "kept_token_count": 737.246875, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9929114878177643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007088501565158367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4940216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.54200439453126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18410, + "valid_token_count": 818.8 + }, + { + "actual_token_mask_ratio": 0.10004719570279122, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.012020138930529356, + "grad_norm": 0.11838701367378235, + "kept_token_count": 736.425, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9938095867633819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006190422642976046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6221496582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.97410278320314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18420, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.100150977820158, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.014370690938085318, + "grad_norm": 0.12112511694431305, + "kept_token_count": 736.4125, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9933168828487396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006683113798499107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2004821777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1299774169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18430, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.10010543167591095, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.0123465521261096, + "grad_norm": 0.09405621886253357, + "kept_token_count": 736.515625, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.992573481798172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007426516246050596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.30066833496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9115020751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18440, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.09931691437959671, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.012738016853109001, + "grad_norm": 0.13568326830863953, + "kept_token_count": 737.08125, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9931031942367554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00689680315554142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.66454162597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9964965820312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18450, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.0998162642121315, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.013952667592093349, + "grad_norm": 0.06744933128356934, + "kept_token_count": 736.565625, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9933233857154846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006676614284515381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9364837646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38747253417966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18460, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.09983174726366997, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.012336105667054653, + "grad_norm": 0.09882896393537521, + "kept_token_count": 736.5890625, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9930025577545166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997430976480245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.78502502441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.849609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18470, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.09976529702544212, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.012409610208123922, + "grad_norm": 0.08910346031188965, + "kept_token_count": 736.8828125, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9934414207935334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006558608170598746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4168243408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9999206542969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18480, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.09985899403691292, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.012662631087005138, + "grad_norm": 0.09598342329263687, + "kept_token_count": 736.4953125, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9930821657180786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006917832791805268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.09624633789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.02049865722654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.09957476556301117, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.013409112486988306, + "grad_norm": 0.09249778091907501, + "kept_token_count": 736.8703125, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9934502840042114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006549710314720869, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.80824584960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.35035400390626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.10009529367089272, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.014153066743165254, + "grad_norm": 0.09857919812202454, + "kept_token_count": 736.5625, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.99320028424263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006799704674631357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8538787841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.37107849121094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18510, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.10000866204500199, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.014211471471935511, + "grad_norm": 0.11249341815710068, + "kept_token_count": 736.7640625, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.993256539106369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006743449997156858, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8374816894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6988800048828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18520, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.10028525590896606, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.011663597030565142, + "grad_norm": 0.09817612916231155, + "kept_token_count": 736.265625, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9935333669185639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064666344784200195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1627685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.44593200683596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18530, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.09975704029202462, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.013668143190443516, + "grad_norm": 0.11405464261770248, + "kept_token_count": 736.8359375, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931426346302032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006857374683022499, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.75263061523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0774719238281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18540, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.10062252432107925, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.012926980061456561, + "grad_norm": 0.10821113735437393, + "kept_token_count": 735.9453125, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9934709250926972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006529067922383547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5310760498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.12218017578124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18550, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.09875202551484108, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.013035338092595338, + "grad_norm": 0.11029133200645447, + "kept_token_count": 737.225, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.993374502658844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006625482905656099, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1523498535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0979705810547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18560, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.10038178265094758, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.01431729383766651, + "grad_norm": 0.11008194833993912, + "kept_token_count": 735.7234375, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992934113740921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007065878435969353, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.06055908203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.5661285400391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18570, + "valid_token_count": 817.8171875 + }, + { + "actual_token_mask_ratio": 0.10011458545923232, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.011711397580802441, + "grad_norm": 0.10095158964395523, + "kept_token_count": 736.590625, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9929437279701233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007056276407092809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.8204772949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.16107177734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18580, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.09971286579966546, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.01388190146535635, + "grad_norm": 0.1023847833275795, + "kept_token_count": 736.7953125, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9932710707187653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006728930212557316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6323516845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.66131896972655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18590, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.10004193484783172, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.012920746766030788, + "grad_norm": 0.14602716267108917, + "kept_token_count": 736.184375, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9935509443283081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449070386588573, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.42918395996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.810791015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18600, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.09948369041085244, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.014145573135465384, + "grad_norm": 0.09744156897068024, + "kept_token_count": 737.0984375, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9934114813804626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00658851619809866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2720916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3657958984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18610, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.1000657819211483, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.012782493233680725, + "grad_norm": 0.09658573567867279, + "kept_token_count": 736.16875, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932750821113586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006724916119128466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.82981262207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.4484588623047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18620, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.10028126090765, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.012755231093615293, + "grad_norm": 0.08672040700912476, + "kept_token_count": 736.4203125, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932124495506287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006787544488906861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8807434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6877197265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18630, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10014954134821892, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.012242525722831488, + "grad_norm": 0.0873740166425705, + "kept_token_count": 736.6078125, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9929641246795654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070358777418732645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.50757751464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.54700622558596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18640, + "valid_token_count": 818.5875 + }, + { + "actual_token_mask_ratio": 0.10032847225666046, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.011661054287105798, + "grad_norm": 0.11118609458208084, + "kept_token_count": 736.1859375, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9933127999305725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006687202863395214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1791564941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.56065673828124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.10018916875123977, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.012618425954133271, + "grad_norm": 0.11042535305023193, + "kept_token_count": 736.415625, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9933139145374298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006686076056212187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.77545166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.93072509765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18660, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.10016237869858742, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.01007025265134871, + "grad_norm": 0.10291449725627899, + "kept_token_count": 736.265625, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9932490408420562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006750973593443632, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9340515136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.39125671386716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18670, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09952737018465996, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.013444128818809986, + "grad_norm": 0.11366871744394302, + "kept_token_count": 737.209375, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9932785511016846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067214460112154486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.42441711425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4048095703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18680, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.09955028146505356, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.013512098602950573, + "grad_norm": 0.11267440021038055, + "kept_token_count": 736.5171875, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9933868527412415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613150890916586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.77662353515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.12950744628904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18690, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.1005216546356678, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.01413312740623951, + "grad_norm": 0.09476854652166367, + "kept_token_count": 735.93125, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9930943369865417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006905654817819596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.88323974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.60874938964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18700, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1004167877137661, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.013151070429012179, + "grad_norm": 0.09989316016435623, + "kept_token_count": 736.2140625, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930343151092529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069656817242503164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.49989929199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3541687011719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18710, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.09962651431560517, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.014151875860989094, + "grad_norm": 0.11843579262495041, + "kept_token_count": 736.5078125, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.993182224035263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00681777885183692, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.62086791992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.45586547851565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18720, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.10014074072241783, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.011701494641602039, + "grad_norm": 0.10465820878744125, + "kept_token_count": 736.515625, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9923780560493469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007621934823691845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7901977539062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9788024902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18730, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.09970221668481827, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.01571883521974087, + "grad_norm": 0.11529054492712021, + "kept_token_count": 737.1671875, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9930664479732514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069335633888840675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.86541442871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8218231201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18740, + "valid_token_count": 818.8015625 + }, + { + "actual_token_mask_ratio": 0.10017935261130333, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.013353702146559954, + "grad_norm": 0.0945369228720665, + "kept_token_count": 736.3296875, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9937385261058808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006261489447206258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.68596801757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9464569091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18750, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.10008376315236092, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.013394815567880868, + "grad_norm": 0.10314786434173584, + "kept_token_count": 736.43125, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9932713747024536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006728635914623737, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3654815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3891845703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18760, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09959867373108863, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.013064667303115129, + "grad_norm": 0.0848456546664238, + "kept_token_count": 736.865625, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9936242818832397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006375706288963556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.83226623535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6033020019531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18770, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09990819692611694, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.012550415936857462, + "grad_norm": 0.08316842466592789, + "kept_token_count": 736.615625, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9934611737728118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065388214774429795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.4718872070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6701629638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18780, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.09989658370614052, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.012120652478188276, + "grad_norm": 0.1047089472413063, + "kept_token_count": 736.4, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9930399835109711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006960006710141897, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9332214355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 403.05194396972655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.09981712922453881, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.01490539163351059, + "grad_norm": 0.10990440845489502, + "kept_token_count": 736.9984375, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.992729902267456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007270096242427826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.0218139648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.44578857421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18800, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.10008664578199386, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.012123755551874638, + "grad_norm": 0.10932818800210953, + "kept_token_count": 736.4453125, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9928389608860015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007161033060401678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.71473083496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2928741455078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18810, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1003050796687603, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.013363689836114644, + "grad_norm": 0.11557812243700027, + "kept_token_count": 736.4484375, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9935940146446228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006405976135283709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1769561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9167419433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18820, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.10015581175684929, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.013159834081307054, + "grad_norm": 0.10753514617681503, + "kept_token_count": 736.1546875, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9935236573219299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006476344913244248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1840057373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.83660888671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18830, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.10016328319907189, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.013279253523796797, + "grad_norm": 0.10118860751390457, + "kept_token_count": 736.2796875, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9933046996593475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006695295497775078, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.40198669433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.39796752929686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18840, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.0999379277229309, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.01411176547408104, + "grad_norm": 0.13609416782855988, + "kept_token_count": 736.6390625, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9931356728076934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006864316761493683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0186248779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.97312927246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18850, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.09994925931096077, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.012455777637660504, + "grad_norm": 0.08646325021982193, + "kept_token_count": 736.2578125, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931649446487427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006835046038031578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2016296386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.41713256835936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18860, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.10001953691244125, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.012483155261725188, + "grad_norm": 0.11573987454175949, + "kept_token_count": 736.53125, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936777532100678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006322245299816132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7326232910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.08262939453124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18870, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.09982845038175583, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.011880132276564837, + "grad_norm": 0.08894651383161545, + "kept_token_count": 736.7046875, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931297957897186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006870204862207174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2326324462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5483001708984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18880, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.10008332878351212, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.012888882169499993, + "grad_norm": 0.09753325581550598, + "kept_token_count": 736.1296875, + "learning_rate": 9.912708818626524e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9937216818332673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278311274945736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.81214294433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0084686279297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18890, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.09986184313893318, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.012331461161375045, + "grad_norm": 0.08076532930135727, + "kept_token_count": 736.6015625, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931978464126587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006802163645625114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4576446533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.2463409423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18900, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.0998438060283661, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.01177163333632052, + "grad_norm": 0.09637438505887985, + "kept_token_count": 736.553125, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9932429611682891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006757025327533484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6370147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9717742919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18910, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.1009636476635933, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.014167182426899672, + "grad_norm": 0.11472911387681961, + "kept_token_count": 735.725, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.993024331331253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006975678820163011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.85988464355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0542022705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18920, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.10019964799284935, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.013969581108540297, + "grad_norm": 0.10477016866207123, + "kept_token_count": 736.24375, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.99280766248703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007192343100905418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9007141113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.45514221191405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18930, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09989535585045814, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.011680604936555029, + "grad_norm": 0.13139212131500244, + "kept_token_count": 736.6, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9929076850414276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092312816530466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.86937255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4577392578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18940, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10036205053329468, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.013194359932094812, + "grad_norm": 0.10553441941738129, + "kept_token_count": 736.1734375, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9928547859191894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007145215757191181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7012939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38929138183596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18950, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.0999688945710659, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.012904172483831643, + "grad_norm": 0.10308372229337692, + "kept_token_count": 736.4296875, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9929921984672546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070077994838356975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1275604248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.23887634277344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18960, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.09956345111131668, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.014971614722162485, + "grad_norm": 0.0720759853720665, + "kept_token_count": 736.528125, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9931406140327453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006859403382986784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.18678588867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.6192352294922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18970, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.10033435449004173, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.013940929947420955, + "grad_norm": 0.1043301448225975, + "kept_token_count": 736.1625, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9934116721153259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588332075625658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.72398681640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.37770690917966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18980, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10055681318044662, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.01332415295764804, + "grad_norm": 0.13918650150299072, + "kept_token_count": 736.1640625, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9929331481456757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007066840771585703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9401885986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.47632751464846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 18990, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.0999754711985588, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.012538838386535644, + "grad_norm": 0.1513563096523285, + "kept_token_count": 736.503125, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9928138315677643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007186154276132584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.96311340332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4607208251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19000, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.10025565773248672, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.012434510048478842, + "grad_norm": 0.10319788753986359, + "kept_token_count": 736.2203125, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993431705236435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006568299699574709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8680755615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.84275512695314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19010, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10011391267180443, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.01204430228099227, + "grad_norm": 0.08118259161710739, + "kept_token_count": 736.1703125, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9931375741958618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006862425338476896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5628326416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9160522460937, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19020, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.1000413492321968, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.014815673837438226, + "grad_norm": 0.12077441811561584, + "kept_token_count": 736.4859375, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0181, + "mgd_cos_sim": 0.9934567034244537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006543308310210705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.31044311523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4378692626953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19030, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09958105608820915, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.015634985268116, + "grad_norm": 0.11428052186965942, + "kept_token_count": 737.1, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0191, + "mgd_cos_sim": 0.9930458068847656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006954187247902155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.48998107910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8728454589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19040, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.09974592179059982, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.013753535551950335, + "grad_norm": 0.08796841651201248, + "kept_token_count": 736.4625, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9929616034030915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038403861224651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0450805664062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9514587402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19050, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.10053635463118553, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.015540637075901031, + "grad_norm": 0.09887103736400604, + "kept_token_count": 735.9328125, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9932648122310639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006735184229910374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6337127685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5691223144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19060, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10012705773115158, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.013908931938931345, + "grad_norm": 0.118134044110775, + "kept_token_count": 736.49375, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9929921388626098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007007853500545025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9704864501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.25384216308595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19070, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.1003687284886837, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.013785966951400042, + "grad_norm": 0.1151118203997612, + "kept_token_count": 736.1359375, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9929907560348511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007009241264313459, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.40615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.18153381347656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19080, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.09944429993629456, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.012655784701928496, + "grad_norm": 0.11882860958576202, + "kept_token_count": 736.7375, + "learning_rate": 9.602876107719228e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932888805866241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006711132358759642, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.73607177734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.82903442382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19090, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10010617151856423, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.01403283029794693, + "grad_norm": 0.13114532828330994, + "kept_token_count": 736.3734375, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9933177053928375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006682307366281748, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9494567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.87567749023435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19100, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.10015227943658829, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.01475654374808073, + "grad_norm": 0.09478802233934402, + "kept_token_count": 736.590625, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9938261926174163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006173795834183693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.59520568847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.45380249023435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19110, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.10029687657952309, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.01474330872297287, + "grad_norm": 0.1277591735124588, + "kept_token_count": 735.8671875, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9930801630020142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006919824052602053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.33363037109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2440551757812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19120, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.09977233707904816, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.013724377192556859, + "grad_norm": 0.13921961188316345, + "kept_token_count": 736.978125, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9930606961250306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939300987869501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.87933349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.5013366699219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19130, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.10012238919734955, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.01437418470159173, + "grad_norm": 0.12111502140760422, + "kept_token_count": 736.109375, + "learning_rate": 9.525821934588551e-06, + "loss": 0.018, + "mgd_cos_sim": 0.9927575409412384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072424645535647866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4224090576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.40806884765624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19140, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.09999679327011109, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.013594552408903837, + "grad_norm": 0.0950201228260994, + "kept_token_count": 736.4, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9933327734470367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006667231116443873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9729064941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7987945556641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19150, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09992926865816117, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.014247099915519357, + "grad_norm": 0.10299312323331833, + "kept_token_count": 736.4390625, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9928281664848327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007171841152012348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.88396911621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.85784912109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19160, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.09967810139060021, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.010736864944919943, + "grad_norm": 0.08785266429185867, + "kept_token_count": 737.09375, + "learning_rate": 9.47966910063128e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933877408504486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00661225700750947, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.19463806152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.98284606933595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19170, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.10025443434715271, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.014051421452313662, + "grad_norm": 0.0906771868467331, + "kept_token_count": 736.3390625, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9934784173965454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00652156574651599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4039825439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8884582519531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19180, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.09980331510305404, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.01363870082423091, + "grad_norm": 0.09250129014253616, + "kept_token_count": 736.790625, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9930887877941131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006911211833357811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8715789794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5902923583984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19190, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.10003584548830986, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.01484802938066423, + "grad_norm": 0.10634855180978775, + "kept_token_count": 736.2859375, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0181, + "mgd_cos_sim": 0.9935429871082306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064570029266178604, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7279571533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.54636840820314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19200, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.10029343366622925, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.012061573285609483, + "grad_norm": 0.11625481396913528, + "kept_token_count": 735.9140625, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9932790696620941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006720935460180044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.47593994140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.65990295410154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19210, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.10053940042853356, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.013132059387862683, + "grad_norm": 0.09274580329656601, + "kept_token_count": 736.128125, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9930021047592164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069979001767933365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4319732666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8823608398437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19220, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.10060248970985412, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.01375507777556777, + "grad_norm": 0.11663953959941864, + "kept_token_count": 735.559375, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9929345071315765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007065499946475029, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.86562194824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6121032714844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19230, + "valid_token_count": 817.8375 + }, + { + "actual_token_mask_ratio": 0.10044243857264519, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.012787992553785443, + "grad_norm": 0.08321542292833328, + "kept_token_count": 735.8390625, + "learning_rate": 9.372214795949655e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935807645320892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00641923863440752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4446960449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.95265808105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19240, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.10043375715613365, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.014270247891545296, + "grad_norm": 0.11650960147380829, + "kept_token_count": 736.1953125, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.993533593416214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00646640108898282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.0111968994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.54538879394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19250, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.09972886145114898, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.01296253316104412, + "grad_norm": 0.08778618276119232, + "kept_token_count": 736.71875, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9934355258941651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006564466934651137, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8321258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7958740234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19260, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.10023911446332931, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.01069129751995206, + "grad_norm": 0.1266893893480301, + "kept_token_count": 736.53125, + "learning_rate": 9.326265152587844e-06, + "loss": 0.014, + "mgd_cos_sim": 0.993472445011139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00652755107730627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.84606018066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3315002441406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19270, + "valid_token_count": 818.584375 + }, + { + "actual_token_mask_ratio": 0.10097446888685227, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.016229243809357287, + "grad_norm": 0.11956892162561417, + "kept_token_count": 735.240625, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0196, + "mgd_cos_sim": 0.993251234292984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006748754531145096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.25023498535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.2588745117188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19280, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.10049219727516175, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.011677823215723037, + "grad_norm": 0.10295423865318298, + "kept_token_count": 735.925, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9930950045585633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006904986966401338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6177947998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.36448059082034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19290, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.10008943378925324, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.014746675919741391, + "grad_norm": 0.11121327430009842, + "kept_token_count": 736.2984375, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9926637887954712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007336202915757895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4480407714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.25709228515626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19300, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.09950278699398041, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.013525023683905601, + "grad_norm": 0.09985555708408356, + "kept_token_count": 737.034375, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9932135462760925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006786472257226705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.18915405273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1304229736328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.09964819103479386, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.01446959814056754, + "grad_norm": 0.10915305465459824, + "kept_token_count": 736.634375, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9933527827262878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006647209264338016, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4022521972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.88258666992186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19320, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.1001557655632496, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.012010336853563786, + "grad_norm": 0.09406611323356628, + "kept_token_count": 736.4078125, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9938295066356659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006170483399182558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.83095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3679504394531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19330, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.10048705935478211, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.012550720665603875, + "grad_norm": 0.12359129637479782, + "kept_token_count": 736.0890625, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934374034404755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00656259311363101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6858367919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.92333068847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19340, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.09977216720581054, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.01114969509653747, + "grad_norm": 0.0774594396352768, + "kept_token_count": 736.6015625, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933829545974732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617056299000978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0995758056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.22980346679685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.09963004663586617, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.012342138774693012, + "grad_norm": 0.09934820234775543, + "kept_token_count": 736.6609375, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9932082056999206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791805755347014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.07383728027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.06133728027345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19360, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.10010202825069428, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.014859777502715587, + "grad_norm": 0.12138985842466354, + "kept_token_count": 736.3109375, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9932699263095855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006730085425078869, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.92315979003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7129760742188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19370, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09927210584282875, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.012546621076762677, + "grad_norm": 0.09019966423511505, + "kept_token_count": 737.1328125, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934337556362152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006566242035478354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.80296936035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.72411193847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19380, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09999926760792732, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.013898313511162996, + "grad_norm": 0.11952192336320877, + "kept_token_count": 736.634375, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9933284819126129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006671502348035574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.59523010253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0862274169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.09998436644673347, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.014133399492129684, + "grad_norm": 0.08956527709960938, + "kept_token_count": 736.6109375, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9934944927692413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006505510210990906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.39918518066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.00559692382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19400, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.09998301714658737, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.012947180029004812, + "grad_norm": 0.12538142502307892, + "kept_token_count": 736.0375, + "learning_rate": 9.112661985779382e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9938962996006012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006103688292205334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6386016845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9652435302734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19410, + "valid_token_count": 817.803125 + }, + { + "actual_token_mask_ratio": 0.1000679925084114, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.013520645536482334, + "grad_norm": 0.13274997472763062, + "kept_token_count": 736.0703125, + "learning_rate": 9.09745766770274e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9930388808250428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006961131189018488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0179412841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.65370788574216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19420, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.10005236938595771, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.011401216872036457, + "grad_norm": 0.15787555277347565, + "kept_token_count": 736.1390625, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937180519104004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006281947158277035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0497772216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8140869140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19430, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.10079242289066315, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.016263695992529394, + "grad_norm": 0.12737508118152618, + "kept_token_count": 735.7109375, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0199, + "mgd_cos_sim": 0.9928199350833893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007180059514939785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2669250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5535034179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19440, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.10095082148909569, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.010036671068519354, + "grad_norm": 0.45764589309692383, + "kept_token_count": 735.71875, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9932354927062989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006764511205255985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9746948242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2511291503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19450, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.10046064928174019, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.012133340816944838, + "grad_norm": 0.08388787508010864, + "kept_token_count": 735.740625, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9934009313583374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006599084381014109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.39658813476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.00791625976564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19460, + "valid_token_count": 817.90625 + }, + { + "actual_token_mask_ratio": 0.09981173425912857, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.012279883539304138, + "grad_norm": 0.11040027439594269, + "kept_token_count": 736.503125, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931792318820953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00682076970115304, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.29172973632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6803283691406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19470, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09999233782291413, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.011705731321126223, + "grad_norm": 0.0986107885837555, + "kept_token_count": 736.2875, + "learning_rate": 9.006383020301024e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9935067117214202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00649328539147973, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.4332305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3049743652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.0994162268936634, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.012141805002465844, + "grad_norm": 0.07578456401824951, + "kept_token_count": 737.275, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9932565927505493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006743395701050758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6930847167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.27862854003905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19490, + "valid_token_count": 818.6609375 + }, + { + "actual_token_mask_ratio": 0.09919192492961884, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.01187677620910108, + "grad_norm": 0.0794517919421196, + "kept_token_count": 737.1015625, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931106388568878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889372412115335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.73912353515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.66484985351565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19500, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.09999614357948303, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.016351563576608896, + "grad_norm": 0.09178124368190765, + "kept_token_count": 736.3578125, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0196, + "mgd_cos_sim": 0.9934054851531983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006594508048146963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.28013610839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4582977294922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19510, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.09976821541786193, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.012860158085823059, + "grad_norm": 0.10983610898256302, + "kept_token_count": 736.3890625, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9930607795715332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939220800995826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.154443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.408642578125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19520, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.10004803165793419, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.014989049686118961, + "grad_norm": 0.11199874430894852, + "kept_token_count": 736.3890625, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0187, + "mgd_cos_sim": 0.9925958514213562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007404165994375944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0690887451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.89556274414065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19530, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.10012408569455147, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.012035241723060608, + "grad_norm": 0.0861186534166336, + "kept_token_count": 736.4859375, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9927569925785065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007243004906922579, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0229156494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.95123596191405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19540, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.10033630952239037, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.013626128993928432, + "grad_norm": 0.10254886001348495, + "kept_token_count": 736.096875, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9934182405471802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006581762526184321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7606658935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38450927734374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19550, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10022006705403327, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.012678401824086905, + "grad_norm": 0.09381850808858871, + "kept_token_count": 736.4328125, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.993073695898056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006926308665424585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4979766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7406768798828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19560, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.09993227943778038, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.013366790162399412, + "grad_norm": 0.09524191170930862, + "kept_token_count": 736.7390625, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.993262505531311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006737491115927696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6905883789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.81456604003904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19570, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.10121183916926384, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.012386464420706034, + "grad_norm": 0.0911397784948349, + "kept_token_count": 735.3734375, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9931916952133178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006808309070765972, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3236602783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1537322998047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19580, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.09943935498595238, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011359542794525623, + "grad_norm": 0.09156937152147293, + "kept_token_count": 737.159375, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936331868171692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366804335266351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6518829345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9275817871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19590, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.10000166147947312, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.013250293117016554, + "grad_norm": 0.12541651725769043, + "kept_token_count": 736.6046875, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9933519661426544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006648038513958454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.26685485839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7495056152344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19600, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.09980895444750786, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.012517694942653179, + "grad_norm": 0.08588043600320816, + "kept_token_count": 736.7046875, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9936154067516327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006384594272822142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4302551269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3116058349609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19610, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.10035404190421104, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.012453799834474921, + "grad_norm": 0.10871969908475876, + "kept_token_count": 736.05, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936289608478546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371053867042065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.057470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0829833984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19620, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10026491060853004, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.01380562297999859, + "grad_norm": 0.10647209733724594, + "kept_token_count": 736.303125, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9934950232505798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006504994258284569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.70753784179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.14775390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19630, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.09972806870937348, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.012505379132926464, + "grad_norm": 0.09946935623884201, + "kept_token_count": 736.6546875, + "learning_rate": 8.764809847932277e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9929469585418701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007053031772375107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6655212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7954040527344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19640, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.10074853524565697, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.013009273447096348, + "grad_norm": 0.10411690920591354, + "kept_token_count": 735.7484375, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9932760775089264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006723937299102545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.76722412109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.5923095703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19650, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.09949801564216613, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.015337745100259781, + "grad_norm": 0.10448597371578217, + "kept_token_count": 736.778125, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9930944085121155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006905587203800678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2044738769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.2472900390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19660, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.10061416700482369, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.015871662367135286, + "grad_norm": 0.1111903265118599, + "kept_token_count": 736.0796875, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0193, + "mgd_cos_sim": 0.9931018233299256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898186821490526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1238647460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6362274169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19670, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.10003410279750824, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.012549581099301576, + "grad_norm": 0.10421206057071686, + "kept_token_count": 736.190625, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933205485343933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006679459102451801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.267626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.48728332519534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19680, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.09962525963783264, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.01241631805896759, + "grad_norm": 0.09008892625570297, + "kept_token_count": 736.9125, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9929986596107483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007001345232129097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.07005615234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 403.5105224609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19690, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.09929998815059662, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.012280953489243985, + "grad_norm": 0.11017853021621704, + "kept_token_count": 736.9625, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.993414968252182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006585039384663105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9877563476563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9700225830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19700, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.0999196469783783, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.010879391944035888, + "grad_norm": 0.07506358623504639, + "kept_token_count": 736.4078125, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9933146476745606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006685353722423315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4650604248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0917144775391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19710, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.10021297559142113, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.012819558195769788, + "grad_norm": 0.08736871182918549, + "kept_token_count": 736.5109375, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9939187109470368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081292312592268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.6249603271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2302978515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19720, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.09977393746376037, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.012041133642196656, + "grad_norm": 0.08968357741832733, + "kept_token_count": 736.353125, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9937071859836578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006292827986180782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.19786376953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6543762207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19730, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.09965794086456299, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.012499868869781494, + "grad_norm": 0.10868272185325623, + "kept_token_count": 736.4203125, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.993772292137146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006227702740579844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9282196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.87339172363284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19740, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.10035703703761101, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.012766440911218525, + "grad_norm": 0.0748680979013443, + "kept_token_count": 736.146875, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9931716978549957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006828308384865522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9711547851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.61900329589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19750, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.09920783787965774, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.015029021771624684, + "grad_norm": 0.10379927605390549, + "kept_token_count": 736.984375, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0186, + "mgd_cos_sim": 0.9928763926029205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007123616151511669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4406005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9583465576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19760, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.10080621093511581, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.011588862864300609, + "grad_norm": 0.10501354932785034, + "kept_token_count": 735.7140625, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9937545537948609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006245446763932705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.53993225097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9989837646484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19770, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.10050286799669265, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.011774517828598619, + "grad_norm": 0.08741496503353119, + "kept_token_count": 735.93125, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9938479840755463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061520240269601345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1886413574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.5424499511719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19780, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09972877204418182, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.014134611189365386, + "grad_norm": 0.07539980113506317, + "kept_token_count": 736.815625, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9933345496654511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006665456295013428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.02916259765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8862365722656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19790, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.09930979087948799, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.013875538529828191, + "grad_norm": 0.0756630152463913, + "kept_token_count": 737.0703125, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9935278832912445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006472114752978086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.57435913085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8361114501953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.09985203295946121, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.01243727058172226, + "grad_norm": 0.07884518802165985, + "kept_token_count": 736.6734375, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9938108742237091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006189134158194065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.30828857421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9650421142578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19810, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1000585064291954, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.011022069770842791, + "grad_norm": 0.07796899229288101, + "kept_token_count": 736.3078125, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934152901172638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00658471267670393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.8365173339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6868530273438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19820, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10030407309532166, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.01356418216601014, + "grad_norm": 0.11387511342763901, + "kept_token_count": 736.20625, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9934162676334382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006583722960203886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.23651733398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.41990966796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19830, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.09913816377520561, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013116235937923193, + "grad_norm": 0.1294812709093094, + "kept_token_count": 736.8546875, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9933011353015899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006698864419013262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.24277954101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1499389648437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19840, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.09972440749406815, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.012478754017502069, + "grad_norm": 0.09798294305801392, + "kept_token_count": 736.6890625, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993250721693039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006749292835593224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7363708496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.36911926269534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.10005302280187607, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.012784588430076838, + "grad_norm": 0.12223448604345322, + "kept_token_count": 736.3578125, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933830022811889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006616995576769114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5517639160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7047821044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10022782981395721, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.012278849119320511, + "grad_norm": 0.0856323093175888, + "kept_token_count": 736.2578125, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9935876548290252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006412333529442549, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.02208251953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.35407104492185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19870, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10035580843687057, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.013886440824717284, + "grad_norm": 0.11668440699577332, + "kept_token_count": 736.359375, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9928873360157013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007112681120634079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.64780578613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.26596984863284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19880, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10049083679914475, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.01274340907111764, + "grad_norm": 0.1088777557015419, + "kept_token_count": 735.890625, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937059462070466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006294071394950152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1539642333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9283782958984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19890, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.09975307136774063, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.011536494083702564, + "grad_norm": 0.0827879011631012, + "kept_token_count": 736.328125, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9935962557792664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064037546515464784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6023376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0297821044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19900, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.10041557103395463, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.014984055235981942, + "grad_norm": 0.10218275338411331, + "kept_token_count": 735.94375, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9921628534793854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007837134599685668, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6470123291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.81370544433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.09956757351756096, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.013113193865865468, + "grad_norm": 0.11968082934617996, + "kept_token_count": 736.5296875, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932025253772736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006797472573816776, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.87205810546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 403.30780029296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19920, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.10127811133861542, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.0174588143825531, + "grad_norm": 0.1277870237827301, + "kept_token_count": 735.39375, + "learning_rate": 8.331953385104516e-06, + "loss": 0.021, + "mgd_cos_sim": 0.99298854470253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070114611648023125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2395751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9282501220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19930, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.0994000867009163, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.011948314681649207, + "grad_norm": 0.08426988124847412, + "kept_token_count": 737.028125, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934138357639313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065861627459526065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.90061645507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.04457397460936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19940, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.09973395764827728, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.012749337777495384, + "grad_norm": 0.10728906840085983, + "kept_token_count": 736.4234375, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9938465654850006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061534462496638295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.319384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.6573211669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19950, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.1002870574593544, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.015639458876103164, + "grad_norm": 0.12404697388410568, + "kept_token_count": 736.190625, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9935216307640076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00647838618606329, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.66437377929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4310760498047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19960, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10088528245687485, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.011232267459854483, + "grad_norm": 0.12335502356290817, + "kept_token_count": 735.684375, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9939824104309082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060175836086273195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.19501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.70575561523435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19970, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.09960166588425637, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.012546253204345704, + "grad_norm": 0.09292666614055634, + "kept_token_count": 736.5578125, + "learning_rate": 8.257999368963374e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930138528347016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00698614502325654, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.98488464355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.23896484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19980, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.09951492473483085, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.01113416813313961, + "grad_norm": 0.08708250522613525, + "kept_token_count": 736.7671875, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9931209146976471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006879085581749678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5824676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8894927978516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 19990, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.09992472156882286, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.013280129246413707, + "grad_norm": 0.09773250669240952, + "kept_token_count": 736.2953125, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9934425413608551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00655745230615139, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0730407714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1856201171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20000, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.0992628961801529, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.010822956776246428, + "grad_norm": 0.12053228914737701, + "kept_token_count": 737.0015625, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9929691553115845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007030842639505863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4492889404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.62965698242186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20010, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.10016295313835144, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.012342458870261908, + "grad_norm": 0.11186657845973969, + "kept_token_count": 736.2015625, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9931582927703857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006841718684881926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8925445556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.92833557128904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20020, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.09998357221484185, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.01401260681450367, + "grad_norm": 0.11710003018379211, + "kept_token_count": 736.3671875, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9933945417404175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006605471856892109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.15234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7960723876953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20030, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.10017532929778099, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.012468080408871174, + "grad_norm": 0.07087909430265427, + "kept_token_count": 736.265625, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934039950370789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065960163250565525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1542510986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6211791992188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20040, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09980511739850044, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.01032693712040782, + "grad_norm": 0.0926373228430748, + "kept_token_count": 736.6984375, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9934952020645141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006504790671169758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9826690673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.53724670410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20050, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.10062393620610237, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.011322544934228063, + "grad_norm": 0.10264214128255844, + "kept_token_count": 735.86875, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931843042373657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006815691757947206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6460662841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.5331665039063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20060, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09919587001204491, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.01266219192184508, + "grad_norm": 0.09363480657339096, + "kept_token_count": 736.984375, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9937590301036835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006240972131490707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4120727539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4762329101562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20070, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.10093913897871971, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.01334099257364869, + "grad_norm": 0.1429992914199829, + "kept_token_count": 735.5203125, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933698117733002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006630189623683691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4780242919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4938629150391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20080, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.09994962215423583, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.01228732792660594, + "grad_norm": 0.06620169430971146, + "kept_token_count": 736.6078125, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9937884330749511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062115656211972235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.9359466552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3784637451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20090, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.10001014620065689, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.011796370847150683, + "grad_norm": 0.09272625297307968, + "kept_token_count": 736.5234375, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9934861361980438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00651385597884655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3455810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9175262451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20100, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.09990514740347863, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.014420348778367042, + "grad_norm": 0.11026149988174438, + "kept_token_count": 736.521875, + "learning_rate": 8.066682997585045e-06, + "loss": 0.018, + "mgd_cos_sim": 0.9928135752677918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007186423987150192, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.82776794433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.82093811035156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10004233196377754, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.012180938757956028, + "grad_norm": 0.1111043393611908, + "kept_token_count": 736.4890625, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933040857315063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006695916969329118, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3032592773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.41383361816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.10041237026453018, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.01278246808797121, + "grad_norm": 0.10514683276414871, + "kept_token_count": 735.946875, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933988809585571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006601114757359028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.54612731933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.47178649902344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20130, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10048983767628669, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.012354810163378716, + "grad_norm": 0.09675668179988861, + "kept_token_count": 736.00625, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993317723274231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006682285573333502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.37085876464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4459991455078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20140, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.1002827189862728, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.012161832116544247, + "grad_norm": 0.07859764993190765, + "kept_token_count": 736.046875, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9934008657932282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006599120143800974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.89166870117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.18670043945315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20150, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.0998708263039589, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.013585013430565595, + "grad_norm": 0.09337963908910751, + "kept_token_count": 736.3890625, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9933417558670044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006658240687102079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.9063659667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0876190185547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20160, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.1005197711288929, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.01383789419196546, + "grad_norm": 0.09435486048460007, + "kept_token_count": 736.34375, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9934207558631897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006579244509339333, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8704895019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.60081481933594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20170, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.10061058774590492, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.011215872876346112, + "grad_norm": 0.08540578186511993, + "kept_token_count": 735.8734375, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9934176445007324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00658235177397728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.302978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6964904785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20180, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.0998222567141056, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.013025253172963857, + "grad_norm": 0.09538798779249191, + "kept_token_count": 736.6921875, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9929736137390137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070263947360217575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8321197509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.92038879394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20190, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.09994921311736107, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.012012086715549231, + "grad_norm": 0.11107863485813141, + "kept_token_count": 736.2390625, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936699092388153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063300970941782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.58905639648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.48142700195314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20200, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.10064231753349304, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.016706517431885004, + "grad_norm": 0.10688763856887817, + "kept_token_count": 735.834375, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0203, + "mgd_cos_sim": 0.9928689241409302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007131076231598854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8356506347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0855255126953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20210, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.10054254233837127, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.01384533690288663, + "grad_norm": 0.10356184840202332, + "kept_token_count": 736.13125, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9930557489395142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006944247987121344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2974090576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5437835693359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20220, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.09949317723512649, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.012796697858721017, + "grad_norm": 0.11471948772668839, + "kept_token_count": 736.8203125, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9931562840938568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006843721680343151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7777435302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5835205078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20230, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.10100571066141129, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.012520295288413762, + "grad_norm": 0.12621285021305084, + "kept_token_count": 735.578125, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9936161398887634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006383866816759109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.0670989990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3070831298828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20240, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.09979300945997238, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.011623026803135872, + "grad_norm": 0.094007208943367, + "kept_token_count": 736.3203125, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.993451988697052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006548013631254434, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.21136474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8188934326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20250, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.10031923353672027, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.012857226561754942, + "grad_norm": 0.0914667546749115, + "kept_token_count": 736.109375, + "learning_rate": 7.847704970163127e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9937475383281708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062524651177227495, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6141815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.49034729003904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20260, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.09990093931555748, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.01265608868561685, + "grad_norm": 0.07903006672859192, + "kept_token_count": 736.5640625, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931284964084626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006871499866247177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2079681396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.93763427734376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20270, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.09997479096055031, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.0119540483225137, + "grad_norm": 0.09102141112089157, + "kept_token_count": 736.4390625, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934831082820892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006516912300139665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.88450927734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5095642089844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20280, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.09920296221971511, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.013215150265023113, + "grad_norm": 0.07993841171264648, + "kept_token_count": 737.0765625, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.993469911813736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006530091911554336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6471343994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0089935302734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20290, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.09941063299775124, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.014005031483247877, + "grad_norm": 0.06733939796686172, + "kept_token_count": 736.9453125, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9929100215435028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007089969608932734, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8628631591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.30723266601564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20300, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.09995529279112816, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.013224445283412933, + "grad_norm": 0.1302875131368637, + "kept_token_count": 736.540625, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931867122650146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006813284289091825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4948486328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.58176574707034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09953379034996032, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.011380058340728283, + "grad_norm": 0.08572141081094742, + "kept_token_count": 736.9484375, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.993676096200943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006323908921331167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.40845947265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0751586914063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10023214966058731, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.01219980833120644, + "grad_norm": 0.0852513462305069, + "kept_token_count": 736.465625, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9937907636165619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062092294916510585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.61158142089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4447509765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20330, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.10027725249528885, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.012793769314885139, + "grad_norm": 0.09769155085086823, + "kept_token_count": 736.41875, + "learning_rate": 7.731712429864325e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935133934020997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006486612092703581, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3415588378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.27333984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20340, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.10027321726083756, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.012666643224656582, + "grad_norm": 0.10315950959920883, + "kept_token_count": 736.1765625, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9930911123752594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006908890139311552, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.570751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3774780273437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.0995640479028225, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.012369397096335888, + "grad_norm": 0.10339637100696564, + "kept_token_count": 736.765625, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9939183592796326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081636436283589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8521331787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.35145568847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20360, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10050484165549278, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.013234126614406705, + "grad_norm": 0.10366421192884445, + "kept_token_count": 735.64375, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9937667787075043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062332327477633955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6774871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0410980224609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20370, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.09962217062711716, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.015428443439304829, + "grad_norm": 0.094239242374897, + "kept_token_count": 736.7671875, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.993300586938858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006699404492974281, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.05979309082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5330291748047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20380, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.09930629283189774, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.011006878828629852, + "grad_norm": 0.1032688319683075, + "kept_token_count": 736.946875, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9936096489429473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006390352919697762, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.16535034179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.745947265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20390, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.09994808956980705, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.013079181127250195, + "grad_norm": 0.10600447654724121, + "kept_token_count": 736.365625, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9935155272483825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00648447722196579, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.896630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3365814208984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20400, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.10022452548146248, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.010990407271310686, + "grad_norm": 0.08181244134902954, + "kept_token_count": 736.2109375, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9932126879692078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067873080261051655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5337249755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.79518432617186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20410, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.09927328154444695, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.01213820707052946, + "grad_norm": 0.12400216609239578, + "kept_token_count": 737.1390625, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935987591743469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006401247251778841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9906707763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.928466796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20420, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.09973259568214417, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.01073860893957317, + "grad_norm": 0.08048880100250244, + "kept_token_count": 736.54375, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9939188182353973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081182975322008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3646575927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.67470397949216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20430, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.09984608143568038, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.013129007816314698, + "grad_norm": 0.11469488590955734, + "kept_token_count": 736.434375, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9934854745864868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006514521222561598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2956878662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7189636230469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20440, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.09962660893797874, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.012320276629179717, + "grad_norm": 0.08071751147508621, + "kept_token_count": 736.840625, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9934800744056702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006519928574562073, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.31414794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8694610595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20450, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.0999574415385723, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.011129553336650133, + "grad_norm": 0.09283766150474548, + "kept_token_count": 736.2421875, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9934939980506897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006506019271910191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.40448608398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6489318847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20460, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.0994363471865654, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.011901519261300564, + "grad_norm": 0.10155855864286423, + "kept_token_count": 736.61875, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931850731372833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006814928539097309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1534057617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9411315917969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20470, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.09994887560606003, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.0159835334867239, + "grad_norm": 0.10340510308742523, + "kept_token_count": 736.4578125, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0192, + "mgd_cos_sim": 0.9936253607273102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637464476749301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9421325683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4715881347656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20480, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.1000982716679573, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.012536088842898608, + "grad_norm": 0.10386160016059875, + "kept_token_count": 736.5, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9929406225681305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007059376128017903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1375793457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.17677917480466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20490, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.10027065873146057, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.010099348798394202, + "grad_norm": 0.09278993308544159, + "kept_token_count": 736.0703125, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9932884037494659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006711598485708237, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.36045532226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7954528808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20500, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.09967324957251548, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.012552930042147636, + "grad_norm": 0.11703073233366013, + "kept_token_count": 736.4265625, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9932299554347992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006770057417452335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0182403564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.28004150390626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20510, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.10056569278240204, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.010659601632505655, + "grad_norm": 0.0793452039361, + "kept_token_count": 735.95625, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9937218606472016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278132647275924, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0334014892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.18194885253905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09950439482927323, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013174732495099307, + "grad_norm": 0.0942617654800415, + "kept_token_count": 736.7453125, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932940065860748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006706009991466999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.80768432617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.25803833007814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20530, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.10041705295443534, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.012885620491579175, + "grad_norm": 0.09759986400604248, + "kept_token_count": 736.06875, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932539522647857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006746051367372274, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2475128173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9640686035156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20540, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10031521916389466, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.012595636257901787, + "grad_norm": 0.14887356758117676, + "kept_token_count": 736.5765625, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9937056064605713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006294398196041584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5111114501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4868621826172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20550, + "valid_token_count": 818.70625 + }, + { + "actual_token_mask_ratio": 0.10009367913007736, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.011437369463965296, + "grad_norm": 0.10015976428985596, + "kept_token_count": 736.328125, + "learning_rate": 7.415674345285763e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9929314851760864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007068514917045831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2426422119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8289794921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20560, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09942628145217895, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.012972245970740915, + "grad_norm": 0.12490902841091156, + "kept_token_count": 737.0359375, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9933643400669098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006635659001767635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1446929931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0269348144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20570, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.10030851811170578, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.013835060317069292, + "grad_norm": 0.09901171177625656, + "kept_token_count": 736.05625, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.993222564458847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006777445785701275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.44362487792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0500823974609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20580, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.09968053102493286, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.01167813278734684, + "grad_norm": 0.07721880823373795, + "kept_token_count": 736.396875, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9940971851348877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00590281430631876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9680114746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4190307617188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20590, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.09971510395407676, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.012502585910260678, + "grad_norm": 0.09948289394378662, + "kept_token_count": 736.5484375, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9936292111873627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637077447026968, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6057861328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4484039306641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20600, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.09964796453714371, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.012701499462127685, + "grad_norm": 0.08636670559644699, + "kept_token_count": 736.84375, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931571543216705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006842836830765009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.16697998046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3583160400391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20610, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.10050468146800995, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.011817308329045772, + "grad_norm": 0.1024240106344223, + "kept_token_count": 736.153125, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9934325635433197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065674371086061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4092315673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.22057189941404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20620, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.0995137333869934, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.013847336173057556, + "grad_norm": 0.09203126281499863, + "kept_token_count": 736.7890625, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9932090997695923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006790891569107771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6039764404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.20147705078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20630, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10063077434897423, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.012509360816329718, + "grad_norm": 0.13310977816581726, + "kept_token_count": 735.775, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993537962436676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064620494842529295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2080352783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6320861816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.10061899945139885, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.011199330352246762, + "grad_norm": 0.09303539246320724, + "kept_token_count": 735.821875, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9934986114501954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065013902261853215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.5679229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0987609863281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20650, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.10066191107034683, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.01238987841643393, + "grad_norm": 0.11436037719249725, + "kept_token_count": 736.2234375, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9929096877574921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007090315874665976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.11119689941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.541650390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20660, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.10028874278068542, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.011203512223437429, + "grad_norm": 0.07206690311431885, + "kept_token_count": 736.509375, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9934477865695953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552223302423954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4009033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.99378967285156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20670, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.10057787597179413, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.013462270656600595, + "grad_norm": 0.0965380072593689, + "kept_token_count": 735.6546875, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9932766079902648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006723381485790015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8900573730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.14696350097654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20680, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.10017043203115464, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.012957930192351341, + "grad_norm": 0.10490849614143372, + "kept_token_count": 736.5765625, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9933803737163543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066196233034133915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.4521148681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.31195373535155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20690, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.09997769370675087, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014069274673238397, + "grad_norm": 0.10180141776800156, + "kept_token_count": 736.5078125, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9928054273128509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00719456747174263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4285614013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.18046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.10038861706852913, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.011619934858754278, + "grad_norm": 0.10286632180213928, + "kept_token_count": 736.178125, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9934587121009827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006541302334517241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.43895568847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.20751037597654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20710, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.1002094529569149, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.014746452029794455, + "grad_norm": 0.09571371227502823, + "kept_token_count": 736.340625, + "learning_rate": 7.188619381260698e-06, + "loss": 0.018, + "mgd_cos_sim": 0.9934310972690582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065688866190612314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5121765136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2600860595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20720, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.10035729482769966, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.012923543341457844, + "grad_norm": 0.07582554221153259, + "kept_token_count": 736.3375, + "learning_rate": 7.174508393396057e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9939384400844574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006061557959765196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0271423339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3160034179688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20730, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.10063719600439072, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.011341821635141969, + "grad_norm": 0.0991058424115181, + "kept_token_count": 735.925, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9939189791679383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060810154303908345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.83286437988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.42262878417966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20740, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10026044249534607, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.011955392919480801, + "grad_norm": 0.11515495926141739, + "kept_token_count": 736.440625, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9937635123729706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006236494611948728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.35932006835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2098724365234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20750, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.09922490939497948, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.010482450155541301, + "grad_norm": 0.11669845879077911, + "kept_token_count": 737.0, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9931954920291901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00680450601503253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.0504974365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0469635009766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20760, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.09985960721969604, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.01239582379348576, + "grad_norm": 0.09744863212108612, + "kept_token_count": 736.628125, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9934497117996216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065502945333719255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.515283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.40241088867185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20770, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10036850571632386, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.012519813422113658, + "grad_norm": 0.10316293686628342, + "kept_token_count": 736.4953125, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993314927816391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006685060635209084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4076721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.07558288574216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20780, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.09974251464009284, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.011909763608127833, + "grad_norm": 0.08867949992418289, + "kept_token_count": 736.4515625, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9936346232891082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006365380063652992, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.9815612792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8728973388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20790, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.10040697753429413, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.010444358503445984, + "grad_norm": 0.09452001005411148, + "kept_token_count": 735.9390625, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9937036097049713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062963954173028466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0945617675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.29493408203126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20800, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.10002490133047104, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.013996225036680698, + "grad_norm": 0.1082063838839531, + "kept_token_count": 736.5875, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9931319177150726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006868076138198375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0812072753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68499145507815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20810, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.09983524680137634, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.01234207432717085, + "grad_norm": 0.10166637599468231, + "kept_token_count": 736.375, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933407247066498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006659287866204977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7295837402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.88329162597654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20820, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.10013057813048362, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.014256935007870198, + "grad_norm": 0.09779252856969833, + "kept_token_count": 736.4859375, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9931752026081085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068247952498495575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.48724975585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5173065185547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.10026588588953018, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.011196833988651633, + "grad_norm": 0.10839121788740158, + "kept_token_count": 736.2171875, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9936802089214325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006319778598845005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.15162658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.68052673339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20840, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10023095235228538, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.01251452681608498, + "grad_norm": 0.11280557513237, + "kept_token_count": 736.2125, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934649229049682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006535086687654257, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3309753417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9543792724609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20850, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09975803494453431, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.012319610081613063, + "grad_norm": 0.14987483620643616, + "kept_token_count": 736.471875, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9939474582672119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006052553467452526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.16535034179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.63328247070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20860, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.09963815733790397, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.012455376842990518, + "grad_norm": 0.09362353384494781, + "kept_token_count": 736.7625, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.993395209312439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066047881729900835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.55322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2046356201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20870, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.09960482567548752, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.010808831918984652, + "grad_norm": 0.08581160753965378, + "kept_token_count": 736.6703125, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9938852965831757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061147032305598256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.05776062011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1404144287109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20880, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.10006259977817536, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.012701934110373259, + "grad_norm": 0.1944076418876648, + "kept_token_count": 736.5078125, + "learning_rate": 6.950037396018328e-06, + "loss": 0.016, + "mgd_cos_sim": 0.993340265750885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006659724377095699, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.19052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0634460449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20890, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.09990370199084282, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.013597959466278554, + "grad_norm": 0.08793438971042633, + "kept_token_count": 736.7875, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9932843148708344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006715701054781675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.25897216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0829345703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20900, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.10025520026683807, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.0129997787065804, + "grad_norm": 0.08610803633928299, + "kept_token_count": 736.246875, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9934897422790527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006510251760482788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1389526367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8598297119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20910, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.0992938905954361, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.013776024337857962, + "grad_norm": 0.09177935868501663, + "kept_token_count": 737.065625, + "learning_rate": 6.908225755978089e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9935443878173829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00645560659468174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2590301513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7999694824219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20920, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.09929628148674965, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.011743992660194635, + "grad_norm": 0.25855398178100586, + "kept_token_count": 737.3578125, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9934440314769745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065559626556932924, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9851104736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.88970642089845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20930, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.10049614459276199, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.013179049780592322, + "grad_norm": 0.12090684473514557, + "kept_token_count": 736.0203125, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9933947265148163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006605271436274052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.557666015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20940, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.09998988956212998, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.012539239134639502, + "grad_norm": 0.09215404093265533, + "kept_token_count": 736.2828125, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9935172200202942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00648276899009943, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.45261535644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.49521484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.10029208809137344, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.013621168397367, + "grad_norm": 0.099860280752182, + "kept_token_count": 736.16875, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9935013055801392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006498687993735075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.69053955078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.57294921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20960, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.10000242963433266, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.013443781156092882, + "grad_norm": 0.14497166872024536, + "kept_token_count": 736.4984375, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9930048704147338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069951342418789865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.12970581054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68463134765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20970, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.10024139657616615, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.013383881375193595, + "grad_norm": 0.1117127314209938, + "kept_token_count": 736.4609375, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.993367338180542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006632662657648325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.90125427246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1219970703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20980, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.09908568412065506, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.012764248996973038, + "grad_norm": 0.10230818390846252, + "kept_token_count": 737.0125, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9938117861747742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006188217177987098, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.60931091308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2196899414063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 20990, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.10009175911545753, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.01116953301243484, + "grad_norm": 0.06839464604854584, + "kept_token_count": 736.290625, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.993764066696167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062359261326491834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2268890380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.58286437988284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21000, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.10071019381284714, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.015073062852025032, + "grad_norm": 0.11379960924386978, + "kept_token_count": 735.8984375, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9936471879482269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006352808140218258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1298522949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.18262634277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.09958042576909065, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.013566712941974401, + "grad_norm": 0.09645011276006699, + "kept_token_count": 736.49375, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.992926561832428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00707342978566885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3451263427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.32715759277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21020, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.10057945549488068, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.012048789905384183, + "grad_norm": 0.10815674066543579, + "kept_token_count": 736.1796875, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9933528482913971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006647145841270685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.98065795898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9632904052734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21030, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.10051962658762932, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.01393073219805956, + "grad_norm": 0.09363679587841034, + "kept_token_count": 735.9859375, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9934239089488983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006576094403862953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7829132080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.70367126464845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21040, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.09994138926267623, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.012693277560174465, + "grad_norm": 0.08412908017635345, + "kept_token_count": 736.5125, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.993274188041687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725814193487167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.198779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2782806396484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21050, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.10055391937494278, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.011408191220834852, + "grad_norm": 0.09284192323684692, + "kept_token_count": 735.978125, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937281012535095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006271899119019509, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.05971069335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3791839599609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21060, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10006014555692673, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.010099040949717164, + "grad_norm": 0.08264458924531937, + "kept_token_count": 736.421875, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9936031758785248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063968307338655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.09654846191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0233520507812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21070, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.09998438879847527, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.01426917603239417, + "grad_norm": 0.09837307780981064, + "kept_token_count": 736.559375, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9927618622779846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007238128874450922, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.9451049804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.66831970214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21080, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.0996578074991703, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.011846736725419759, + "grad_norm": 0.09586764872074127, + "kept_token_count": 736.709375, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.993587076663971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006412917096167803, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6431518554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7725372314453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21090, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.10029512345790863, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.012322331359609962, + "grad_norm": 0.12343702465295792, + "kept_token_count": 736.2625, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9935128033161164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006487214751541614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.58172607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.57843627929685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21100, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.10016312897205353, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.011926077026873826, + "grad_norm": 0.09324046969413757, + "kept_token_count": 736.16875, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9935111343860626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006488882936537266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.643212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1459686279297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21110, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.10069919377565384, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.012558110663667322, + "grad_norm": 0.0995863825082779, + "kept_token_count": 736.01875, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9930101990699768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006989793758839369, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9532440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5961212158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21120, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.09982598274946212, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.013486297335475683, + "grad_norm": 0.1137532964348793, + "kept_token_count": 736.7265625, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9934311628341674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006568838842213154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.77557373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.07769165039065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21130, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.09968999102711677, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.011599746719002723, + "grad_norm": 0.08877439051866531, + "kept_token_count": 736.4296875, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933697581291199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006630246527493, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7605834960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6863677978516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21140, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.10110305994749069, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.012080036848783494, + "grad_norm": 0.0815492644906044, + "kept_token_count": 735.365625, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.993682724237442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317272316664457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3109497070312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.53935546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21150, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.09886730685830117, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.010739638563245535, + "grad_norm": 0.08962835371494293, + "kept_token_count": 737.203125, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9936255097389222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006374483555555343, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.01031494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4092102050781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09995083957910537, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.01262645199894905, + "grad_norm": 0.09949174523353577, + "kept_token_count": 736.7984375, + "learning_rate": 6.563279940911124e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9933353722095489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006664626114070415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.58632507324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.87109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21170, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.10003107115626335, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.012343505397439003, + "grad_norm": 0.10263194143772125, + "kept_token_count": 736.553125, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993260657787323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006739338580518961, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6253356933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0829132080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.10024924576282501, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.013801663089543581, + "grad_norm": 0.1380814164876938, + "kept_token_count": 736.075, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9934794008731842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520593073219061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.50929260253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5804748535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21190, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09916949942708016, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.01175171104259789, + "grad_norm": 0.07983346283435822, + "kept_token_count": 737.3078125, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9932278275489808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006772181857377291, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3507781982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.342138671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21200, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.10043270662426948, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.012519263383001088, + "grad_norm": 0.0973237082362175, + "kept_token_count": 735.9359375, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931559860706329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006844013836234808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.87232360839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.2071228027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21210, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.1002229742705822, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.010474418941885232, + "grad_norm": 0.10457473248243332, + "kept_token_count": 736.2546875, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9932544589042663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067455340176820755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.0381713867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.0248962402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10042331218719483, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.01328437919728458, + "grad_norm": 0.10782522708177567, + "kept_token_count": 735.8328125, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9934072911739349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006592706684023142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 460.46171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2616790771484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21230, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.09972387105226517, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.012465315172448754, + "grad_norm": 0.12354717403650284, + "kept_token_count": 736.8, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9938164412975311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006183558609336615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6657440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.55992431640624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21240, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.0995862454175949, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.010239244112744928, + "grad_norm": 0.10381709039211273, + "kept_token_count": 736.775, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935163855552673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006483616400510072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9724639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9909423828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21250, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.0999116063117981, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.010601234110072254, + "grad_norm": 0.10246578603982925, + "kept_token_count": 736.434375, + "learning_rate": 6.440654282174203e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9932021558284759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006797854509204626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9388061523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6267333984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21260, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.10016668736934661, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.011587264947593212, + "grad_norm": 0.08979924023151398, + "kept_token_count": 736.0859375, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9936125993728637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006387406215071678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.36858825683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0275848388672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21270, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.09889148250222206, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.011392186256125569, + "grad_norm": 0.10288491100072861, + "kept_token_count": 737.246875, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9939183712005615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00608163932338357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4850219726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8255676269531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21280, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.10032353922724724, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.011692388728260995, + "grad_norm": 0.07451863586902618, + "kept_token_count": 736.3203125, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9937662541866302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006233752053231001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.48935546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.10792541503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21290, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.10013676807284355, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.012119216984137893, + "grad_norm": 0.11191920936107635, + "kept_token_count": 736.384375, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935735106468201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006426494847983122, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2835296630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3650726318359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21300, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.10046505853533745, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.01409499989822507, + "grad_norm": 0.11489936709403992, + "kept_token_count": 736.0875, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9935372948646546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006462708208709955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4637145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2166809082031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21310, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.0993297502398491, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.010796136967837811, + "grad_norm": 0.1003674566745758, + "kept_token_count": 737.0109375, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934358179569245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006564178224653006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6560974121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5719421386719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21320, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09951745569705964, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.011571782547980547, + "grad_norm": 0.08616814762353897, + "kept_token_count": 736.921875, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9938106060028076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006189390737563372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9165313720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.61949768066404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21330, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.10036170333623887, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.01206376114860177, + "grad_norm": 0.10900573432445526, + "kept_token_count": 735.8546875, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9934613287448884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006538659986108542, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8488525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 403.24835510253905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21340, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.10008179396390915, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.014097871910780668, + "grad_norm": 0.11957304179668427, + "kept_token_count": 736.3515625, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9929717302322387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007028279360383749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.01988525390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.46412353515626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.10016092136502266, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.01256888904608786, + "grad_norm": 0.1309584677219391, + "kept_token_count": 736.14375, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9937179267406464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628205705434084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2244049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6144256591797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21360, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.09998193681240082, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.010707153147086501, + "grad_norm": 0.10665334761142731, + "kept_token_count": 736.3640625, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993042266368866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006957745552062989, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9160522460937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1090942382813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21370, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.09937841296195984, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.010971077810972929, + "grad_norm": 0.11133341491222382, + "kept_token_count": 736.9921875, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934245705604553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065754251554608345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.05186462402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8151489257813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21380, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.10026219934225082, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.011345264548435808, + "grad_norm": 0.10816513001918793, + "kept_token_count": 736.0890625, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9932552099227905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006744798272848129, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.21213989257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0309326171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.09975041300058365, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.012790311314165591, + "grad_norm": 0.11120231449604034, + "kept_token_count": 736.5734375, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9923949062824249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007605096511542797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.73085021972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7031585693359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21400, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.10009729862213135, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.012294512707740068, + "grad_norm": 0.07956309616565704, + "kept_token_count": 736.1734375, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9933218836784363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006678120233118534, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5846374511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.21629943847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.09986591041088104, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.012390529131516814, + "grad_norm": 0.09729602932929993, + "kept_token_count": 736.3328125, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9936959207057953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006304070539772511, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1570220947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4541961669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21420, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.09918132796883583, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.011979729821905493, + "grad_norm": 0.09995529055595398, + "kept_token_count": 737.1671875, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993214362859726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067856419831514355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7213439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2326934814453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21430, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.10051952600479126, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.01220285277813673, + "grad_norm": 0.1282838135957718, + "kept_token_count": 735.9625, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933793723583222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006620630063116551, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4213928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3545440673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21440, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10071595013141632, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.01064481120556593, + "grad_norm": 0.08217667788267136, + "kept_token_count": 735.99375, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9938604414463044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006139567773789168, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4903900146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.712744140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21450, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.09980027675628662, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.011963789770379662, + "grad_norm": 0.11003623157739639, + "kept_token_count": 736.6078125, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9937458574771881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006254149414598942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5043914794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7674224853516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21460, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.10099765658378601, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.01297557344660163, + "grad_norm": 0.08796600997447968, + "kept_token_count": 735.6859375, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9935143709182739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006485645566135645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.16095275878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8826416015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21470, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.09988416954874993, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.012528163194656373, + "grad_norm": 0.08955029398202896, + "kept_token_count": 736.4859375, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9933574855327606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642518378794193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.64129638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.18393859863284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21480, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.10019686371088028, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.012199126370251178, + "grad_norm": 0.08915658295154572, + "kept_token_count": 736.034375, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9937249898910523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006275004614144564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1494506835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3617370605469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21490, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.10031107887625694, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.01131504406221211, + "grad_norm": 0.07514471560716629, + "kept_token_count": 736.1125, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936632990837098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006336703430861235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.16734924316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9733062744141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21500, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.09996118694543839, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.012270324816927313, + "grad_norm": 0.11610100418329239, + "kept_token_count": 736.525, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9937500715255737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00624992772936821, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8013977050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.83167724609376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21510, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.10034133195877075, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.013363803550601006, + "grad_norm": 0.11327750980854034, + "kept_token_count": 736.2375, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933242619037628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006675747223198414, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.91001892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.24445190429685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21520, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.1005866751074791, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011197644053027034, + "grad_norm": 0.13537536561489105, + "kept_token_count": 736.2390625, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9938090682029724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006190921738743782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.09796447753905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21530, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.09996639639139175, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.011671508010476828, + "grad_norm": 0.13487176597118378, + "kept_token_count": 736.565625, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9927625358104706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007237465959042311, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.63844299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.19652099609374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21540, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.09991474747657776, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.012608707323670388, + "grad_norm": 0.10070652514696121, + "kept_token_count": 736.4109375, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9934062600135803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065937436185777186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.29877319335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6006256103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21550, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.10009419322013854, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.01195134692825377, + "grad_norm": 0.09780123084783554, + "kept_token_count": 736.5484375, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.993628853559494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371139548718929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.48316650390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.00733642578126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21560, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.099374920129776, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.012158188689500093, + "grad_norm": 0.12985287606716156, + "kept_token_count": 737.025, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933443069458008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00665570218116045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.79381103515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.56585083007815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21570, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.0999742865562439, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.0124661133158952, + "grad_norm": 0.09797387570142746, + "kept_token_count": 736.515625, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9939383804798126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006061618309468031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8923736572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2980438232422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21580, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.10025113224983215, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.011429965170100332, + "grad_norm": 0.09459478408098221, + "kept_token_count": 736.3203125, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937900245189667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062099639326334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1339904785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.05527648925784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21590, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.10006983280181884, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.011958227679133415, + "grad_norm": 0.14277788996696472, + "kept_token_count": 736.490625, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9936580896377564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341910269111395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.03346557617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6607727050781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21600, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.10021292045712471, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.012717635743319989, + "grad_norm": 0.10138482600450516, + "kept_token_count": 735.9515625, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9932163298130036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006783659569919109, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.91798706054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.0465789794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21610, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.10046245008707047, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.012118066009134055, + "grad_norm": 0.10407641530036926, + "kept_token_count": 736.1703125, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935454368591309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00645456463098526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9821350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6065399169922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21620, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.10036675184965134, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.01170569183304906, + "grad_norm": 0.10338058322668076, + "kept_token_count": 736.33125, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9936889827251434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00631100544705987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.01589050292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6934020996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21630, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.1003434456884861, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.010930012306198477, + "grad_norm": 0.09236826747655869, + "kept_token_count": 735.971875, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9934906959533691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006509302277117967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.84649658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4011291503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21640, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.0998196005821228, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.012296814331784844, + "grad_norm": 0.07431812584400177, + "kept_token_count": 736.8828125, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9933529615402221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006647041253745556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.38194580078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.55540771484374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21650, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.0998434342443943, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.010388048505410552, + "grad_norm": 0.08990433812141418, + "kept_token_count": 736.6125, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9940961956977844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00590380672365427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6382202148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.52894287109376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21660, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.09982362240552903, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.012630187347531319, + "grad_norm": 0.10452540963888168, + "kept_token_count": 736.1875, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9936163544654846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006383656989783049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3650268554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3356414794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21670, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.09932845085859299, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.013148754928261041, + "grad_norm": 0.09014369547367096, + "kept_token_count": 737.1015625, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932898759841919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00671014217659831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.09352111816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.6189727783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21680, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.10015847459435463, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.01005398272536695, + "grad_norm": 0.11059632897377014, + "kept_token_count": 736.484375, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9932897448539734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00671025337651372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4550048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.972265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21690, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.09951358363032341, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.01312275193631649, + "grad_norm": 0.08872651308774948, + "kept_token_count": 736.959375, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9934922456741333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006507749110460282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.63177795410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3423614501953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21700, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.09936361163854598, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.011713487189263105, + "grad_norm": 0.08524703234434128, + "kept_token_count": 736.921875, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931194663047791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006880534440279007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.44187622070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.2105377197266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.09959838986396789, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.01269489168189466, + "grad_norm": 0.11235092580318451, + "kept_token_count": 736.825, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9936392068862915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006360793486237526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5811279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.50694580078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21720, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.10017562732100486, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.012105339393019676, + "grad_norm": 0.11559740453958511, + "kept_token_count": 735.9859375, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9935081720352172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006491824518889189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.35198364257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4329345703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21730, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.10010529309511185, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.010327454656362534, + "grad_norm": 0.06694115698337555, + "kept_token_count": 736.340625, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.993740838766098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006259168125689029, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2929382324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2616760253906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21740, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.10044675394892692, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.013975445227697492, + "grad_norm": 0.16228841245174408, + "kept_token_count": 735.8359375, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.993578439950943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006421566009521484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.60535583496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6720672607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21750, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.10014661401510239, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.011822867859154939, + "grad_norm": 0.07965761423110962, + "kept_token_count": 736.4421875, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9932404696941376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00675953347235918, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9679412841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5965270996094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21760, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.10102701112627983, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.011548790987581015, + "grad_norm": 0.14349666237831116, + "kept_token_count": 735.6, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9937528133392334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247190572321415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0893127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3542785644531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21770, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09944032654166221, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.013153593335300685, + "grad_norm": 0.10890050232410431, + "kept_token_count": 736.865625, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9935177505016327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006482247635722161, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0315826416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1331512451172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21780, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09996773526072503, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.009677687706425787, + "grad_norm": 0.07497848570346832, + "kept_token_count": 736.5140625, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939041316509247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006095869466662407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.63025817871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2050415039063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.10003960877656937, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.012096344027668238, + "grad_norm": 0.09794452786445618, + "kept_token_count": 736.30625, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9934028387069702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006597159430384636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.73545227050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8060607910156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21800, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.09916131272912025, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.013157673180103302, + "grad_norm": 0.11555274575948715, + "kept_token_count": 737.19375, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9934849202632904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006515073031187058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.505712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8684356689453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.09995197728276253, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.012183728255331516, + "grad_norm": 0.10040681064128876, + "kept_token_count": 736.6390625, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993482255935669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006517743971198797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6263061523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.75132141113284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21820, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.10018982514739036, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.012731037754565477, + "grad_norm": 0.09022179991006851, + "kept_token_count": 736.0703125, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937396049499512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260401010513306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.85703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4012390136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21830, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.09977805837988854, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.011319357203319669, + "grad_norm": 0.12020745873451233, + "kept_token_count": 736.4296875, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935529232025146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006447082385420799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5611602783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7450775146484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21840, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.09933495298027992, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.01285978383384645, + "grad_norm": 0.09118091315031052, + "kept_token_count": 737.1515625, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993900579214096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006099423114210367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9160186767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.34142150878904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21850, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.1000357449054718, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.012647249596193432, + "grad_norm": 0.1017487496137619, + "kept_token_count": 736.3, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9935052752494812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006494721118360758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.03692626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.70274353027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21860, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.09966895431280136, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.012769321817904711, + "grad_norm": 0.10051602870225906, + "kept_token_count": 736.846875, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935677945613861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064321991987526415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.08359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6633728027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21870, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.10004887357354164, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.012730511464178562, + "grad_norm": 0.11123188585042953, + "kept_token_count": 736.40625, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993586927652359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006413076352328062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5268585205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.48203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10025302097201347, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.012559543224051594, + "grad_norm": 0.1475161761045456, + "kept_token_count": 736.5390625, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.993514609336853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006485390849411488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2139129638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.03350219726565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21890, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.10056222602725029, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.011144012585282326, + "grad_norm": 0.06401260942220688, + "kept_token_count": 736.0546875, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933823049068451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617692206054926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7889801025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4641082763672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21900, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.09970795139670371, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.012076313840225339, + "grad_norm": 0.1241842582821846, + "kept_token_count": 736.684375, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9937413394451141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258669029921293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.68226013183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.53539123535154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21910, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09974675476551056, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.01139361597597599, + "grad_norm": 0.11678533256053925, + "kept_token_count": 736.6375, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9933856368064881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006614358723163604, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.55299682617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7742980957031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.09994065389037132, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.014143112860620023, + "grad_norm": 0.10150300711393356, + "kept_token_count": 736.6015625, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9934625208377839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537471152842045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6365081787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.12250061035155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21930, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.10002694427967071, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.013092719297856093, + "grad_norm": 0.1256614774465561, + "kept_token_count": 736.5046875, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932170033454895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006783007364720106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.952685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68040466308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21940, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10006714463233948, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.012201726529747247, + "grad_norm": 0.11594817042350769, + "kept_token_count": 736.63125, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9934800505638123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065199657343328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6867340087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.36205139160154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21950, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.10007601007819175, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.01184686361812055, + "grad_norm": 0.1023041233420372, + "kept_token_count": 736.4359375, + "learning_rate": 5.516570963231058e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9936962962150574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006303707882761955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.270947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0491638183594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21960, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.10004713833332061, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.01183840548619628, + "grad_norm": 0.08124794811010361, + "kept_token_count": 736.4921875, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9938779294490814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006122072599828244, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.99558715820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5977416992188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21970, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.09903650060296058, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.013255731062963605, + "grad_norm": 0.08151955157518387, + "kept_token_count": 737.1984375, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9929366052150727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00706338956952095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1870941162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.13865051269534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.10018298923969268, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.011841284250840545, + "grad_norm": 0.0996132493019104, + "kept_token_count": 736.03125, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933651864528656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006634818203747272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.52940673828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.24224243164065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 21990, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.09969187155365944, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.012510885065421463, + "grad_norm": 0.11169914901256561, + "kept_token_count": 736.6203125, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.993427449464798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006572563573718071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7100433349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8268096923828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22000, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.10053829029202462, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.011798288067802786, + "grad_norm": 0.12138301134109497, + "kept_token_count": 736.0125, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931949138641357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006805109139531851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7444732666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5343963623047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22010, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.10111364349722862, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.01245083617977798, + "grad_norm": 0.12333609163761139, + "kept_token_count": 735.5375, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936726212501525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006327386945486069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6341033935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0031341552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22020, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09990137815475464, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.011474219243973493, + "grad_norm": 0.13670890033245087, + "kept_token_count": 736.3375, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9937741041183472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225908454507589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.12871704101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9519989013672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22030, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.09998194575309753, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.011900530569255352, + "grad_norm": 0.1888919621706009, + "kept_token_count": 736.434375, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934504806995392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006549521628767252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.964794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.45294189453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.10054570585489273, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.012258680257946252, + "grad_norm": 0.08285193145275116, + "kept_token_count": 735.96875, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9938839733600616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006116008479148149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.579052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5911590576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.099844691157341, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.00913037327118218, + "grad_norm": 0.09014951437711716, + "kept_token_count": 736.3921875, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936710298061371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006328981928527355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.82887573242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4725402832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22060, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.09993357434868813, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.011768650216981768, + "grad_norm": 0.07948838174343109, + "kept_token_count": 736.275, + "learning_rate": 5.37635293136319e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9935064733028411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493525486439467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.23056945800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9478820800781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22070, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.09917540252208709, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.012088643014431, + "grad_norm": 0.07998545467853546, + "kept_token_count": 737.259375, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935684204101562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006431582756340504, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.51585083007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1343475341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22080, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.09994450733065605, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.010675286827608942, + "grad_norm": 0.1306418627500534, + "kept_token_count": 736.490625, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9939847409725189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006015261355787516, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5768188476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.38995971679685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22090, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10011744201183319, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.011628000671043992, + "grad_norm": 0.07664286345243454, + "kept_token_count": 736.5640625, + "learning_rate": 5.338356350951308e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9933324158191681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006667596939951182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.38088073730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.382373046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22100, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.10002672970294953, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.01292850123718381, + "grad_norm": 0.11117362976074219, + "kept_token_count": 736.7046875, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932694733142853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006730536092072725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.91465148925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4919677734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22110, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.09986612796783448, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.009970849985256791, + "grad_norm": 0.10364513099193573, + "kept_token_count": 736.559375, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935778021812439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006422189064323902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4852996826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22120, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.1006153367459774, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.013075352692976595, + "grad_norm": 0.08258602023124695, + "kept_token_count": 736.375, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9935751914978027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006424807570874691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.87525939941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.94512329101565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22130, + "valid_token_count": 818.75625 + }, + { + "actual_token_mask_ratio": 0.09931812211871147, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.011591993598267436, + "grad_norm": 0.10266510397195816, + "kept_token_count": 736.8859375, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9934998333454133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00650017186999321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7255889892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0775451660156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22140, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.10006474256515503, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.01136844684369862, + "grad_norm": 0.10848011076450348, + "kept_token_count": 736.1578125, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9933788061141968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066212015226483345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3551361083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.604443359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22150, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.10032160207629204, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.014876136276870965, + "grad_norm": 0.07342910766601562, + "kept_token_count": 736.33125, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0181, + "mgd_cos_sim": 0.9935024976730347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006497485190629959, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3773468017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.04192810058595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22160, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.1000486433506012, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.011859213747084141, + "grad_norm": 0.11124901473522186, + "kept_token_count": 736.4140625, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9939655065536499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006034500896930695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3528594970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9171966552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22170, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.10048637241125107, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.011414519837126135, + "grad_norm": 0.08671607822179794, + "kept_token_count": 735.846875, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937424540519715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257536448538303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.12428588867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.19801025390626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22180, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.09994971677660942, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.01062053102068603, + "grad_norm": 0.08035411685705185, + "kept_token_count": 736.296875, + "learning_rate": 5.225002215153973e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9932390809059143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006760911270976067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6361511230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.31658325195315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.09987974688410758, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.012015519244596363, + "grad_norm": 0.10969752073287964, + "kept_token_count": 736.4703125, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9937212944030762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278698612004518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1795135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.05201721191406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22200, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.09955950304865838, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.014189612213522195, + "grad_norm": 0.11912811547517776, + "kept_token_count": 736.8390625, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9931900262832641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00680996123701334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2359222412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.100732421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22210, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09984228238463402, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.010298146586865186, + "grad_norm": 0.09211811423301697, + "kept_token_count": 736.575, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9940118968486786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005988103710114956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4150085449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.63160095214846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22220, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.0995005838572979, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.011014656396582723, + "grad_norm": 0.10324996709823608, + "kept_token_count": 736.653125, + "learning_rate": 5.174930753146763e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9940976798534393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005902325734496117, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6707427978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4740905761719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.10016088485717774, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.01066326373256743, + "grad_norm": 0.09902352839708328, + "kept_token_count": 736.44375, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9934543251991272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545664742588997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9412414550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4512939453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22240, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.10005121082067489, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.01090915915556252, + "grad_norm": 0.08395320922136307, + "kept_token_count": 736.2484375, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9940560519695282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059439513832330705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7549713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.76044921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22250, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.10019644498825073, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.01092096627689898, + "grad_norm": 0.11207841336727142, + "kept_token_count": 736.4046875, + "learning_rate": 5.137502456402705e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938581645488739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006141840200871229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5127899169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2606140136719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22260, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.09988434463739396, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.012365484330803155, + "grad_norm": 0.10345610976219177, + "kept_token_count": 736.415625, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933808326721192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006619175337255001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.04750671386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7944793701172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22270, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.1000988468527794, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.013056781655177475, + "grad_norm": 0.1283445507287979, + "kept_token_count": 736.0828125, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9930405735969543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006959417089819908, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4961456298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0583068847656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22280, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.10012650862336159, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.010703790606930852, + "grad_norm": 0.08438290655612946, + "kept_token_count": 736.415625, + "learning_rate": 5.10018201435167e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934245109558105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006575482431799173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.27928161621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5579406738281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22290, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09973979964852334, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.013649460114538669, + "grad_norm": 0.08632481098175049, + "kept_token_count": 736.5375, + "learning_rate": 5.087765905286004e-06, + "loss": 0.017, + "mgd_cos_sim": 0.993218743801117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006781240832060576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2805572509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1041748046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22300, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.09985889643430709, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.011773895844817162, + "grad_norm": 0.08176200836896896, + "kept_token_count": 736.615625, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9935296773910522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006470341049134731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.41419677734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7875091552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.10004860758781434, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.011725077126175164, + "grad_norm": 0.1090802550315857, + "kept_token_count": 736.31875, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933043539524078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006695642042905093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2242462158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1068511962891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22320, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.10088647529482841, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013363973004743457, + "grad_norm": 0.07194007933139801, + "kept_token_count": 735.9515625, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9931926608085633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006807347387075424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5500518798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8712585449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22330, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.09982969611883163, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.012414028868079185, + "grad_norm": 0.16474205255508423, + "kept_token_count": 736.20625, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936810433864594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063189415261149405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3892456054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.478857421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22340, + "valid_token_count": 817.8546875 + }, + { + "actual_token_mask_ratio": 0.10029587000608445, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.013354943180456758, + "grad_norm": 0.12543249130249023, + "kept_token_count": 736.3453125, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9936626017093658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00633739372715354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8414337158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3989288330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.1000888928771019, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.010423810314387083, + "grad_norm": 0.08158811181783676, + "kept_token_count": 736.240625, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9935304641723632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006469533126801252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5947540283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.16371154785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22360, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10018562003970147, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.012763563636690378, + "grad_norm": 0.10451450943946838, + "kept_token_count": 736.1421875, + "learning_rate": 5.001191225212096e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935969650745392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00640304833650589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9916229248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.31243286132815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22370, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.09933813735842704, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.011531673837453127, + "grad_norm": 0.10829747468233109, + "kept_token_count": 736.7203125, + "learning_rate": 4.9888718917363e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9930544674396515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006945543736219406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.90041198730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.530810546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22380, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.09999010115861892, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.013018700340762735, + "grad_norm": 0.0966922789812088, + "kept_token_count": 736.85, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9940008103847504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059991925023496154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.83396606445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0414672851563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22390, + "valid_token_count": 818.7109375 + }, + { + "actual_token_mask_ratio": 0.10012404471635819, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.012207981571555138, + "grad_norm": 0.10487351566553116, + "kept_token_count": 736.6140625, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9932744443416596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725556403398514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3537200927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6142272949219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22400, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.10035232082009315, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.012964173685759306, + "grad_norm": 0.09444541484117508, + "kept_token_count": 735.8015625, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9937278687953949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006272125896066428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9646270751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8606262207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22410, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.10071968287229538, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.013328430615365506, + "grad_norm": 0.0811752900481224, + "kept_token_count": 736.23125, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9936573565006256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063426470384001735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5477783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5476989746094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22420, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.09977615550160408, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.011357292206957937, + "grad_norm": 0.09946787357330322, + "kept_token_count": 736.828125, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9935149729251862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006485026609152555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7555267333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0619873046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22430, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.10037161633372307, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.012581178080290556, + "grad_norm": 0.11550609022378922, + "kept_token_count": 736.0671875, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9937848031520844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006215184926986694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2089477539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.95355224609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22440, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10002982839941979, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.013343527633696794, + "grad_norm": 0.07729250192642212, + "kept_token_count": 736.3375, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9937193036079407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006280701514333487, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0562683105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1936492919922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22450, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.10028406754136085, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.013138050679117441, + "grad_norm": 0.06623812764883041, + "kept_token_count": 736.2328125, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9939287602901459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006071244832128286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7610626220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6854217529297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.09911431446671486, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010029663098976016, + "grad_norm": 0.09043876081705093, + "kept_token_count": 737.0359375, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9936812937259674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006318691559135914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9355895996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8325225830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22470, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.09992006272077561, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.010532515309751033, + "grad_norm": 0.0904417484998703, + "kept_token_count": 736.5015625, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.993581759929657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064182493835687636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.62388916015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.52752075195315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22480, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.10004845783114433, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.009955900721251965, + "grad_norm": 0.0964028537273407, + "kept_token_count": 736.45, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9936848521232605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006315152253955602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0832550048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5639190673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22490, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.10046192556619644, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.011308998335152864, + "grad_norm": 0.10446666926145554, + "kept_token_count": 736.0265625, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937245666980743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062754198908805845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.98670959472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6591552734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22500, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.0999703899025917, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.012421618681401014, + "grad_norm": 0.10769034922122955, + "kept_token_count": 736.3046875, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9934682369232177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531766336411238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99956970214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.5876831054687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22510, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.10009962692856789, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013370271306484938, + "grad_norm": 0.1153644397854805, + "kept_token_count": 736.3703125, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933986306190491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006601364444941282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.71392211914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7310729980469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22520, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.0999681681394577, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.009714198252186178, + "grad_norm": 0.0975804477930069, + "kept_token_count": 736.471875, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9933189034461976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006681101303547621, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.87656860351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9003540039063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09917503818869591, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.012522503593936562, + "grad_norm": 0.09642874449491501, + "kept_token_count": 736.7140625, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9937654912471772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006234501302242279, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6858825683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0402435302734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22540, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.09964141026139259, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.0120039704721421, + "grad_norm": 0.08221320807933807, + "kept_token_count": 736.34375, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9934027135372162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006597286742180586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6975524902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9803741455078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22550, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.09954841360449791, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.010308671602979302, + "grad_norm": 0.0850248783826828, + "kept_token_count": 736.7796875, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937268078327179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006273189559578896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9632843017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.71649475097655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22560, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10099739283323288, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.011289279814809561, + "grad_norm": 0.08943687379360199, + "kept_token_count": 735.5375, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933901429176331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006609866302460432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.65186767578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.11159057617186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09968693628907203, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.012680153269320727, + "grad_norm": 0.11050571501255035, + "kept_token_count": 736.909375, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.992960798740387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007039206195622682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.21190490722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.56944580078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22580, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.09932118356227874, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.011615872010588646, + "grad_norm": 0.11218620836734772, + "kept_token_count": 736.834375, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9935924589633942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006407533306628466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0731597900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.10455322265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22590, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.10031970515847206, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.010565008595585823, + "grad_norm": 0.10248329490423203, + "kept_token_count": 736.1109375, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.993768447637558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006231547798961401, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.44542236328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8232055664063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22600, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09957159236073494, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.013412266597151756, + "grad_norm": 0.09305422753095627, + "kept_token_count": 737.0390625, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9930597901344299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006940203625708819, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.87574157714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.64121704101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22610, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.09991351291537284, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.013111339136958122, + "grad_norm": 0.0960705429315567, + "kept_token_count": 736.3703125, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932043075561523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006795691512525082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5136657714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.47801818847654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22620, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.10047545582056046, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.012509450037032365, + "grad_norm": 0.0797162726521492, + "kept_token_count": 736.2578125, + "learning_rate": 4.684880664776199e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9929344654083252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007065537758171559, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2843444824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8542022705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22630, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.10020604580640793, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.010708939470350742, + "grad_norm": 0.09873189777135849, + "kept_token_count": 736.11875, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9931704461574554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006829553842544555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.32005615234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4703674316406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22640, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.10004920810461045, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.012572852708399297, + "grad_norm": 0.08969686925411224, + "kept_token_count": 736.009375, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933700561523438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006629951391369104, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2443328857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.66786804199216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22650, + "valid_token_count": 817.83125 + }, + { + "actual_token_mask_ratio": 0.10033479928970337, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.012749272910878063, + "grad_norm": 0.1138211116194725, + "kept_token_count": 736.2359375, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937397181987763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260283291339874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8246704101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.90712280273436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22660, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.10128771215677261, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.012968549132347107, + "grad_norm": 0.08569300919771194, + "kept_token_count": 735.4578125, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9929293632507324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007070639450103044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.66082153320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7715118408203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22670, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.09988851174712181, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.01156697324477136, + "grad_norm": 0.1086765006184578, + "kept_token_count": 736.5765625, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933710396289825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066289685666561125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2312408447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38707275390624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22680, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.100162772834301, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.013745516818016768, + "grad_norm": 0.09149447083473206, + "kept_token_count": 736.3875, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9933848202228546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006615166179835796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9938201904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.95850830078126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22690, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.09993838667869567, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.012169537646695972, + "grad_norm": 0.1008162796497345, + "kept_token_count": 736.31875, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9932941555976867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006705851946026087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.1584167480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2795928955078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22700, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.0997500516474247, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.013298446871340274, + "grad_norm": 0.0998774990439415, + "kept_token_count": 736.46875, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9933969616889954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006603038590401411, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.125244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.97425231933596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22710, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.10080288723111153, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.013011231366544963, + "grad_norm": 0.10483600199222565, + "kept_token_count": 735.9515625, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.993164336681366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00683565428480506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.83665466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7942840576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22720, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.1000008873641491, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.01037601362913847, + "grad_norm": 0.08684277534484863, + "kept_token_count": 736.44375, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9931990206241608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006800989340990782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.32912292480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2831268310547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22730, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09967872276902198, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.013435568427667022, + "grad_norm": 0.08803379535675049, + "kept_token_count": 736.41875, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9939192354679107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006080756150186062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8374755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0549560546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22740, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.1006468839943409, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.012400903226807713, + "grad_norm": 0.1045813038945198, + "kept_token_count": 736.0125, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9930261433124542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006973858270794153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9253326416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.02614135742186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22750, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.09975822940468788, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012421638611704111, + "grad_norm": 0.08780160546302795, + "kept_token_count": 736.515625, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936420440673828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006357953883707523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2132202148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.44745483398435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22760, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.10055625662207604, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.011955682653933764, + "grad_norm": 0.08074788004159927, + "kept_token_count": 735.9140625, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9936181545257569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381840724498033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.70745849609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4926727294922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.09940070658922195, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.01252674893476069, + "grad_norm": 0.11047860980033875, + "kept_token_count": 736.95625, + "learning_rate": 4.506231315070235e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930890738964081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00691090701147914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.19068908691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5175018310547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.10072613060474396, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.011817935667932034, + "grad_norm": 0.08470863848924637, + "kept_token_count": 735.628125, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9940167963504791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005983197875320911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.93866271972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3302398681641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22790, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.10085169896483422, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.009656054945662618, + "grad_norm": 0.08248423784971237, + "kept_token_count": 735.9703125, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939265191555023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006073480378836394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.80242919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.13759460449216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22800, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.10017069429159164, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.011304301861673593, + "grad_norm": 0.10266434401273727, + "kept_token_count": 736.2140625, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937616646289825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062383214943110945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5425720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.80878295898435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22810, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.10025528073310852, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.010006706835702062, + "grad_norm": 0.06305450946092606, + "kept_token_count": 736.2703125, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933665633201599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006633430533111095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.22442321777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.06817321777345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22820, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.09973096251487731, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.009427862567827105, + "grad_norm": 0.06207665428519249, + "kept_token_count": 737.04375, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.993886798620224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006113191787153483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3094512939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2063354492187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22830, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.10011157765984535, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.012071990640833974, + "grad_norm": 0.13751570880413055, + "kept_token_count": 736.0953125, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9931454956531525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006854509934782982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.16290588378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.721826171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22840, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.10035300478339196, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.011052522389218212, + "grad_norm": 0.1079031452536583, + "kept_token_count": 736.23125, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9932622373104095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006737759988754988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.66968994140626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.10036136284470558, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.011298255482688545, + "grad_norm": 0.06331431120634079, + "kept_token_count": 736.3140625, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935000479221344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006499962043017149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4979217529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4048675537109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22860, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.0998989313840866, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.013740526419132948, + "grad_norm": 0.11590588837862015, + "kept_token_count": 736.1625, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9928927481174469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007107248902320862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.24635925292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.24017333984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22870, + "valid_token_count": 817.8671875 + }, + { + "actual_token_mask_ratio": 0.10045384243130684, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.012824878143146635, + "grad_norm": 0.06666070222854614, + "kept_token_count": 736.0046875, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993808650970459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006191354431211949, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.01533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9710418701172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22880, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10000013485550881, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.012161902571097016, + "grad_norm": 0.12824232876300812, + "kept_token_count": 736.18125, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9935749709606171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064250238239765166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.487744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5570953369141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22890, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.09949975833296776, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.011308363312855363, + "grad_norm": 0.05656019225716591, + "kept_token_count": 736.821875, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936812877655029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006318712327629328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.73586120605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6741485595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22900, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.10020459070801735, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.011607079347595573, + "grad_norm": 0.09187650680541992, + "kept_token_count": 736.2375, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9939931631088257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060068369843065735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0366668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.50780029296874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.09985935688018799, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.012282783444970847, + "grad_norm": 0.0898190438747406, + "kept_token_count": 736.346875, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9937030017375946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062969991005957125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.87567443847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.89032287597655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22920, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.10051672831177712, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.012818561121821403, + "grad_norm": 0.07897970825433731, + "kept_token_count": 736.065625, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9935288429260254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006471156980842352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5171813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.39765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22930, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.09967522025108337, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.011889976356178522, + "grad_norm": 0.1472035050392151, + "kept_token_count": 736.6734375, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934042930603028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006595714390277863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.66368713378904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7051177978516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22940, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.09977316781878472, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.010221050214022398, + "grad_norm": 0.09325490891933441, + "kept_token_count": 736.6296875, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9939263820648193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006073623243719339, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.46376647949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.46026000976565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10029465928673745, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.01009174007922411, + "grad_norm": 0.09034614264965057, + "kept_token_count": 736.1140625, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9937164604663848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006283546332269907, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.67213439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.90455017089846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22960, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.10011503919959068, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.01184755926951766, + "grad_norm": 0.13757164776325226, + "kept_token_count": 736.428125, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933014810085297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006698517687618732, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4319244384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6983703613281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22970, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.10025679916143418, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.012156086321920157, + "grad_norm": 0.0827125683426857, + "kept_token_count": 736.89375, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9934956014156342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006504395604133606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.21674194335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6657196044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22980, + "valid_token_count": 819.003125 + }, + { + "actual_token_mask_ratio": 0.10048220008611679, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.013160359067842364, + "grad_norm": 0.137870192527771, + "kept_token_count": 736.1703125, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9935210943222046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006478895433247089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.16246643066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.74237670898435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 22990, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.10021211206912994, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.011203360138460993, + "grad_norm": 0.08836742490530014, + "kept_token_count": 736.20625, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9936684250831604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006331568397581577, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.73890991210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1263671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23000, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.09933229684829711, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.011417425656691193, + "grad_norm": 0.13561242818832397, + "kept_token_count": 736.8734375, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.993179726600647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006820277683436871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0730377197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4152526855469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23010, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.1011979565024376, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.011301817558705806, + "grad_norm": 0.08512607961893082, + "kept_token_count": 735.40625, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936014890670777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063985145650804045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8576141357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2216033935547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23020, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10040592402219772, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.012681238772347569, + "grad_norm": 0.09593522548675537, + "kept_token_count": 735.9859375, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9935301423072815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064698544330894945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8977111816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7490264892578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23030, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.09915350675582886, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.012453881977126002, + "grad_norm": 0.09773983061313629, + "kept_token_count": 737.1625, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9938006460666656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006199347879737615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4184173583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.73968505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23040, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.10060088261961937, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.013980684662237763, + "grad_norm": 0.11847030371427536, + "kept_token_count": 735.9484375, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9933606088161469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639391928911209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.115087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.71595764160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23050, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.09994010627269745, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.011293404316529632, + "grad_norm": 0.10116078704595566, + "kept_token_count": 736.659375, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933407545089722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066592440009117125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.97247314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1146728515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23060, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.09941729828715325, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.01140917232260108, + "grad_norm": 0.09243194013834, + "kept_token_count": 737.19375, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9933232069015503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006676793098449707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1571746826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9972259521484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23070, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.10048280581831932, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.012036209739744664, + "grad_norm": 0.09501362591981888, + "kept_token_count": 736.1796875, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.993096262216568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006903739273548126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5891418457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.342724609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23080, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.10031940639019013, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.01143821612931788, + "grad_norm": 0.09349881857633591, + "kept_token_count": 736.165625, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9934740900993347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006525913905352354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6344390869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.17327575683595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23090, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.09961696267127991, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.012431115470826625, + "grad_norm": 0.10261151194572449, + "kept_token_count": 736.8859375, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9937014281749725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006298563536256551, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8250793457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3930999755859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23100, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.09982779547572136, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.01278028143569827, + "grad_norm": 0.09982798248529434, + "kept_token_count": 736.7671875, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9936620831489563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006337917130440474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.06513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3251983642578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23110, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.09977570623159408, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.011001827428117394, + "grad_norm": 0.07952336221933365, + "kept_token_count": 736.6390625, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935747981071472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006425210554152727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.76001892089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.29694213867185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23120, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.09973099157214164, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.010810569021850824, + "grad_norm": 0.099085234105587, + "kept_token_count": 736.7921875, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937216222286225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278382800519467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.91294250488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1339538574219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23130, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.09935901984572411, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.010325598763301969, + "grad_norm": 0.09166604280471802, + "kept_token_count": 737.1828125, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9933381021022797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006661889515817166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.95794677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.87998962402344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23140, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.10038561001420021, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.013429120462387801, + "grad_norm": 0.11932071298360825, + "kept_token_count": 736.3046875, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9936893224716187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006310677900910377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5502075195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6275177001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.0994712844491005, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.013610128220170737, + "grad_norm": 0.11959145218133926, + "kept_token_count": 736.9171875, + "learning_rate": 4.066602533087398e-06, + "loss": 0.017, + "mgd_cos_sim": 0.993244343996048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006755667924880982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4315643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.16863403320315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23160, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.10008633956313133, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.01040542502887547, + "grad_norm": 0.08447015285491943, + "kept_token_count": 735.9984375, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9935749173164368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064250772818923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3567596435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.00437622070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23170, + "valid_token_count": 817.85625 + }, + { + "actual_token_mask_ratio": 0.09943892061710358, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.011671323049813508, + "grad_norm": 0.07131770253181458, + "kept_token_count": 737.053125, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9937489628791809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006251047737896443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7217163085937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.19642333984376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23180, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09969383925199508, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.012729955930262804, + "grad_norm": 0.09720049053430557, + "kept_token_count": 737.0171875, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9935633480548859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006436643935739994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7761627197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6960418701172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23190, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.10074032843112946, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.010829296475276352, + "grad_norm": 0.11336742341518402, + "kept_token_count": 736.0109375, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934882581233978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006511731073260307, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6315032958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0487457275391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.09998571947216987, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.01029042094014585, + "grad_norm": 0.09600341320037842, + "kept_token_count": 736.759375, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937847018241882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006215296499431133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1553131103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.42503967285154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23210, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.10037243962287903, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.012309834640473128, + "grad_norm": 0.09267907589673996, + "kept_token_count": 736.1453125, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9938464939594269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061534993350505825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2246337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9320037841797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23220, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.09987381771206856, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.011039565596729518, + "grad_norm": 0.09420806169509888, + "kept_token_count": 736.3765625, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9938503503799438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006149646732956171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1454345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5600616455078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23230, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.10009306222200394, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.012345509417355061, + "grad_norm": 0.09906946122646332, + "kept_token_count": 736.553125, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9936432123184205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006356789637356997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5882568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8690979003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23240, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.10029607787728309, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.011448810901492834, + "grad_norm": 0.06635057926177979, + "kept_token_count": 736.1234375, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9926718175411224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007328183855861425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.7935729980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1906402587891, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23250, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.09944986701011657, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.014106637658551335, + "grad_norm": 0.07317022234201431, + "kept_token_count": 736.784375, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9932624936103821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006737521942704916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8866302490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.84226379394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23260, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.10026667267084122, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.010142012871801854, + "grad_norm": 0.07544469088315964, + "kept_token_count": 736.2046875, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9934669196605682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533075496554375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8483551025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3472961425781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23270, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10019179582595825, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.012451095320284367, + "grad_norm": 0.07758783549070358, + "kept_token_count": 736.1859375, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931884348392487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00681155351921916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8421203613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9502777099609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.10004454329609871, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.010503195226192474, + "grad_norm": 0.08711503446102142, + "kept_token_count": 736.634375, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9935364186763763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006463567819446325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3932586669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.170654296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23290, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.09995271936058998, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.011199001548811793, + "grad_norm": 0.09314744919538498, + "kept_token_count": 736.5140625, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9939908027648926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006009211577475071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8665344238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3151519775391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23300, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.10006482601165771, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.011789373494684697, + "grad_norm": 0.07351444661617279, + "kept_token_count": 736.48125, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9938161134719848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006183876376599074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7369384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4681457519531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23310, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10074726715683938, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.00967059782706201, + "grad_norm": 0.0944901779294014, + "kept_token_count": 735.990625, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939318120479583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006068189535290003, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1070526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1000732421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23320, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.09987824857234955, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.010614442871883512, + "grad_norm": 0.0746142715215683, + "kept_token_count": 736.5296875, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9934200048446655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006580001395195723, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.15563049316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1972930908203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23330, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09966531991958619, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.013458899920806289, + "grad_norm": 0.11168235540390015, + "kept_token_count": 736.6171875, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9934804677963257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006519538536667824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.17832641601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0990844726563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23340, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.10046456009149551, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.01076793079264462, + "grad_norm": 0.12477728724479675, + "kept_token_count": 736.0203125, + "learning_rate": 3.853934072822881e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9935622155666352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006437782105058432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.95025329589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.01336975097655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09962530136108398, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.011654645856469869, + "grad_norm": 0.09188169986009598, + "kept_token_count": 736.4875, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9934197962284088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006580204982310533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1810577392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.11204528808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23360, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.10005558729171753, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.013041788153350354, + "grad_norm": 0.11033642292022705, + "kept_token_count": 736.3578125, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9930346250534058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006965379323810339, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.73377685546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2888458251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23370, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.1000425361096859, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.01173083670437336, + "grad_norm": 0.07062335312366486, + "kept_token_count": 736.4921875, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9939840018749238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006016001850366592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.36118469238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6873291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23380, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10004762932658195, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.01139236968010664, + "grad_norm": 0.09611152112483978, + "kept_token_count": 736.103125, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9931525230407715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006847478915005922, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9039672851562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23390, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.09947818517684937, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.01081025912426412, + "grad_norm": 0.10720556229352951, + "kept_token_count": 736.8921875, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9939511120319366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060488985851407055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.54249877929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2786499023438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23400, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09986837282776832, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.01135111073963344, + "grad_norm": 0.07044827193021774, + "kept_token_count": 736.5265625, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936480104923249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006351975165307522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4922210693359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0033294677734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.09946802631020546, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.012822398077696561, + "grad_norm": 0.07092425972223282, + "kept_token_count": 737.11875, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9938368678092957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006163128092885018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.01878662109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7353912353516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23420, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.09957580864429474, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.012181390216574073, + "grad_norm": 0.06600815802812576, + "kept_token_count": 736.596875, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9936360359191895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00636396361514926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.63465270996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8110717773437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23430, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.1000322550535202, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.011284743808209896, + "grad_norm": 0.08891469240188599, + "kept_token_count": 736.121875, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9938796818256378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006120315007865429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.811962890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7859252929687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23440, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.09996306151151657, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.010700724925845862, + "grad_norm": 0.10934215039014816, + "kept_token_count": 736.6140625, + "learning_rate": 3.743961212594197e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933871567249298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612842623144388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1254638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0903350830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23450, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.10018955692648887, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.013663308788090944, + "grad_norm": 0.09261073172092438, + "kept_token_count": 736.1484375, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9935490190982819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006450985930860043, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.51979370117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.24452819824216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23460, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.09982349798083305, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.013391821971163153, + "grad_norm": 0.067182257771492, + "kept_token_count": 736.415625, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9938145756721497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006185417901724577, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7497192382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9329559326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23470, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.09997348636388778, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.012662395648658276, + "grad_norm": 0.08682424575090408, + "kept_token_count": 736.4859375, + "learning_rate": 3.711235452582775e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9933025419712067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006697468459606171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.720556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.894970703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.09941793382167816, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.010896798223257065, + "grad_norm": 0.07486400008201599, + "kept_token_count": 736.9125, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9933086514472962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006691349484026432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.98320922851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38104248046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23490, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.10025776252150535, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.012459488492459059, + "grad_norm": 0.07819908857345581, + "kept_token_count": 736.09375, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9937645196914673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006235478539019823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.58762512207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3470489501953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23500, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.09985306113958359, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.0111651002895087, + "grad_norm": 0.08491919934749603, + "kept_token_count": 736.4953125, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993913447856903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060865446925163266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7870208740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9130157470703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23510, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.10048711597919464, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.012552923429757357, + "grad_norm": 0.13061639666557312, + "kept_token_count": 736.21875, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9940167546272278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005983232986181974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2392211914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6099334716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23520, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.0996182106435299, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.012823820812627673, + "grad_norm": 0.07945001870393753, + "kept_token_count": 736.96875, + "learning_rate": 3.656967033397544e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9936254680156708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006374517641961575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.00014038085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6641418457031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23530, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.10009575560688973, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.01285504000261426, + "grad_norm": 0.10384450852870941, + "kept_token_count": 736.80625, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9939946472644806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006005355715751648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.58319091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1830749511719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23540, + "valid_token_count": 818.759375 + }, + { + "actual_token_mask_ratio": 0.10002283379435539, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.012094316724687814, + "grad_norm": 0.08443617075681686, + "kept_token_count": 736.484375, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.993683522939682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006316477991640567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4476379394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9219177246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23550, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.09977810308337212, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.012420338671654462, + "grad_norm": 0.11810596287250519, + "kept_token_count": 736.3484375, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936986148357392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006301380228251219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8260467529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.52948608398435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23560, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.10007772296667099, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.012848011311143636, + "grad_norm": 0.10389439761638641, + "kept_token_count": 736.625, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9938522160053254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006147767789661885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3734130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4680603027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23570, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.09989942163228989, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.011361514125019312, + "grad_norm": 0.11100324988365173, + "kept_token_count": 736.6546875, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9939324676990509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006067541334778071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1748565673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.76023254394534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23580, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.10007696822285653, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.011478487309068442, + "grad_norm": 0.08125525712966919, + "kept_token_count": 736.4703125, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.99344482421875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006555186957120896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.66522827148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.8548553466797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23590, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.09965084418654442, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.011721539311110973, + "grad_norm": 0.07295122742652893, + "kept_token_count": 736.7640625, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9936345875263214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006365410052239895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.87237548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9409973144531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23600, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.09953884705901146, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.011406541289761662, + "grad_norm": 0.0953494980931282, + "kept_token_count": 736.625, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9936314344406127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063685659319162365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.63782653808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.17091369628906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23610, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.10023018196225167, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.013445958402007819, + "grad_norm": 0.10046300292015076, + "kept_token_count": 736.175, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9936331450939179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366866733878851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.16009216308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0420715332031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23620, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.09986099973320961, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.010954997269436717, + "grad_norm": 0.10419412702322006, + "kept_token_count": 736.796875, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938975393772125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006102464441210031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7156616210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.75653381347655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23630, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.09983175173401833, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.010774738667532802, + "grad_norm": 0.10785402357578278, + "kept_token_count": 736.5765625, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937744855880737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225505378097296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4232116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.579150390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10036680996417999, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.011499960673972964, + "grad_norm": 0.09099709987640381, + "kept_token_count": 736.071875, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9936418056488037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006358174234628677, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0314697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4768890380859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23650, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09969418197870254, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.01398014985024929, + "grad_norm": 0.09217340499162674, + "kept_token_count": 736.628125, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9940859377384186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005914057325571776, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8457977294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.47511291503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23660, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.09969954192638397, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.01107319318689406, + "grad_norm": 0.08150940388441086, + "kept_token_count": 736.7984375, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937458097934723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006254194304347038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.60760192871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3522644042969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23670, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.09906512126326561, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.010157338669523596, + "grad_norm": 0.08998127281665802, + "kept_token_count": 737.10625, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9932181656360626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006781853456050158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7245666503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.02806396484374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23680, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.10045929551124573, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.011271116090938449, + "grad_norm": 0.0897403135895729, + "kept_token_count": 736.1953125, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935881078243256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064119015820324424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.27263793945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.09194030761716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23690, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.0998194083571434, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.010826828004792333, + "grad_norm": 0.09161269664764404, + "kept_token_count": 736.3890625, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9935500919818878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449911929666996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9737152099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2688232421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23700, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.10013632252812385, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.010992152290418744, + "grad_norm": 0.09392911940813065, + "kept_token_count": 736.1109375, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9939439475536347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006056036613881588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.582568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9014556884766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23710, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.10000304505228996, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.01195091400295496, + "grad_norm": 0.09447518736124039, + "kept_token_count": 736.4359375, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9935272812843323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006472718529403209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6545074462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.8720977783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23720, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.09972657710313797, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.011980264540761709, + "grad_norm": 0.08401969075202942, + "kept_token_count": 736.5890625, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.993523758649826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006476236879825592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3083190917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.06495666503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23730, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.09989420622587204, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.011656962148845196, + "grad_norm": 0.08670693635940552, + "kept_token_count": 736.6640625, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.993812108039856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006187896616756916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.74229736328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.72954711914065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23740, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.09954674392938614, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.013066531019285321, + "grad_norm": 0.09205206483602524, + "kept_token_count": 737.0578125, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.99340460896492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00659537510946393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.79462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2743896484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23750, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.09981851950287819, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.012841572333127261, + "grad_norm": 0.09823410958051682, + "kept_token_count": 736.58125, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937994420528412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006200554221868515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.23158874511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7783599853516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10025706440210343, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.011008351994678377, + "grad_norm": 0.09298708289861679, + "kept_token_count": 736.196875, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9931313991546631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006868583150207996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.524169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6763458251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23770, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.10011963844299317, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.011264941515401005, + "grad_norm": 0.10128681361675262, + "kept_token_count": 736.33125, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.993926465511322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006073528900742531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.31141967773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.87694091796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23780, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10100854113698006, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.011298959609121084, + "grad_norm": 0.08784791827201843, + "kept_token_count": 735.4890625, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934445142745971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065554861910641195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1946166992187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3852874755859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23790, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.10041294246912003, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.01128888470120728, + "grad_norm": 0.09216596931219101, + "kept_token_count": 735.9125, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937802612781524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006219728477299213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.14409790039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7868194580078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23800, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.1002421498298645, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.012499095825478434, + "grad_norm": 0.12215711921453476, + "kept_token_count": 736.390625, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934546053409576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545388605445623, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.25733337402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7960540771484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23810, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.10002067759633064, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.011827325634658336, + "grad_norm": 0.09344973415136337, + "kept_token_count": 736.315625, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9935116529464721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00648834565654397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4159698486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0965972900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23820, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.09904428571462631, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.013747736904770136, + "grad_norm": 0.10693196952342987, + "kept_token_count": 737.2453125, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9936526954174042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00634731026366353, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9506378173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.86525573730466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23830, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.10010296031832695, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.011812180746346712, + "grad_norm": 0.10027458518743515, + "kept_token_count": 736.5359375, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931903183460236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006809688918292522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6526153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6168151855469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23840, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.09956290796399117, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.013088933285325765, + "grad_norm": 0.1489865630865097, + "kept_token_count": 736.7890625, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932727098464966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067272957414388655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.71843872070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 402.240771484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23850, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.09961508139967919, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012109524058178067, + "grad_norm": 0.09385553747415543, + "kept_token_count": 736.634375, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935809910297394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006419010087847709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.85805053710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.57965698242185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23860, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.09984311237931251, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.010533000901341438, + "grad_norm": 0.10848549008369446, + "kept_token_count": 736.74375, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9939473152160645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006052668578922749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0881378173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2674194335938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23870, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.0995855487883091, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.010098150698468089, + "grad_norm": 0.08768095076084137, + "kept_token_count": 736.7953125, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.993540471792221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006459522061049938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9622375488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.85308227539065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23880, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.10024840980768204, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.010889965994283556, + "grad_norm": 0.09414783120155334, + "kept_token_count": 736.5109375, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9935701787471771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006429820321500301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.30665588378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.65114440917966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23890, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.09985562562942504, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.011915282625705004, + "grad_norm": 0.08645231276750565, + "kept_token_count": 736.565625, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.994067645072937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005932358559221029, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.97100219726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.12810974121095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23900, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.10021613165736198, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.01086909705772996, + "grad_norm": 0.07428856194019318, + "kept_token_count": 736.51875, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9941451489925385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005854860506951809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.7476043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6333465576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23910, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.09997162818908692, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.009551942022517324, + "grad_norm": 0.09989013522863388, + "kept_token_count": 736.790625, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9936846733093262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006315312627702951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.00209655761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1313140869141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23920, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.1002335049211979, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.011310052080079912, + "grad_norm": 0.10706627368927002, + "kept_token_count": 736.328125, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9935150563716888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064849377609789375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.15191345214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.96376953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23930, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.09975890666246415, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.011583852861076594, + "grad_norm": 0.08950213342905045, + "kept_token_count": 736.7921875, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9934628129005432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537178531289101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2319732666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.53423461914065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23940, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.10040087923407555, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.01262088273651898, + "grad_norm": 0.12029217183589935, + "kept_token_count": 736.403125, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9934719026088714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006528107728809118, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.77080993652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0116394042969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23950, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.09983079433441162, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.0137445826549083, + "grad_norm": 0.10272300988435745, + "kept_token_count": 736.89375, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9934689462184906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531056482344866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7205383300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.24257507324216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23960, + "valid_token_count": 818.61875 + }, + { + "actual_token_mask_ratio": 0.09980237931013107, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.01287049581296742, + "grad_norm": 0.2742014229297638, + "kept_token_count": 736.5171875, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9934579491615295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065420612692832945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4979583740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2528869628906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.0994299903512001, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.011058715870603919, + "grad_norm": 0.09620219469070435, + "kept_token_count": 736.8796875, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9939049303531646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006095078215003014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8773986816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.20872192382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23980, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.10012120828032493, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.011528068594634533, + "grad_norm": 0.10662270337343216, + "kept_token_count": 736.4578125, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9936169028282166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006383105553686619, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7316162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1149139404297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 23990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.099614829570055, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.010621372330933809, + "grad_norm": 0.0986681580543518, + "kept_token_count": 736.7546875, + "learning_rate": 3.163908013150124e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933341324329377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006665864679962397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.49712829589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.56475219726565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24000, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10028086379170417, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.012975861923769117, + "grad_norm": 0.06705071032047272, + "kept_token_count": 735.9515625, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9936813712120056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063186299055814745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1052947998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.00624694824216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24010, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.10010280087590218, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.011804599221795798, + "grad_norm": 0.0684938058257103, + "kept_token_count": 736.46875, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9936883449554443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311659142374993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6440856933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.32305603027345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24020, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10004077404737473, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.010745247593149543, + "grad_norm": 0.1531044840812683, + "kept_token_count": 736.5234375, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9936423659324646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063576363027095795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8474945068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6284606933594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24030, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.10029254704713822, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.009311508340761065, + "grad_norm": 0.10000325739383698, + "kept_token_count": 736.1, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9939018428325653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006098152883350849, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9693664550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.08937683105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24040, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10050788298249244, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.011793647101148963, + "grad_norm": 0.14003023505210876, + "kept_token_count": 736.1078125, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933506429195404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006649362575262785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6234069824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.92828063964845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24050, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.10017080083489419, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.011074723210185766, + "grad_norm": 0.12609858810901642, + "kept_token_count": 736.075, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937238454818725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062761446461081505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8608642578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.84043579101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24060, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.1000061236321926, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.011749117728322744, + "grad_norm": 0.09624424576759338, + "kept_token_count": 736.4265625, + "learning_rate": 3.09316025464495e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9934198975563049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006580115016549826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6660400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.57151489257814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24070, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.09971214607357978, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.011388506414368749, + "grad_norm": 0.11131346225738525, + "kept_token_count": 736.853125, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9941610217094421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005838985275477171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5819396972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7544464111328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.10032978802919387, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.011112211691215634, + "grad_norm": 0.09155402332544327, + "kept_token_count": 735.9125, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935817003250123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064182977192103864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99127197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8224395751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24090, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.09975568279623985, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.011724385991692543, + "grad_norm": 0.13208027184009552, + "kept_token_count": 736.3328125, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9938244640827179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006175548769533634, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5642395019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.95429077148435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24100, + "valid_token_count": 817.921875 + }, + { + "actual_token_mask_ratio": 0.10018584504723549, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.01015523886308074, + "grad_norm": 0.07647601515054703, + "kept_token_count": 736.4046875, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9937288105487824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062712011858820915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4084533691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0117248535156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24110, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.09996907711029053, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.010230922186747194, + "grad_norm": 0.12896394729614258, + "kept_token_count": 736.1625, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9932561039924621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006743891444057226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7777923583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3476257324219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24120, + "valid_token_count": 817.93125 + }, + { + "actual_token_mask_ratio": 0.1004073090851307, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.012124348804354668, + "grad_norm": 0.10383255034685135, + "kept_token_count": 736.0140625, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9936433076858521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063566937111318115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8538513183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.82318725585935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24130, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.09980429410934448, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.010314731393009424, + "grad_norm": 0.10479871928691864, + "kept_token_count": 736.621875, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9930704653263092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006929534766823053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2143859863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5493103027344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24140, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.10000866428017616, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.010453752195462584, + "grad_norm": 0.07020921260118484, + "kept_token_count": 736.25, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9936135172843933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006386502180248499, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.925, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.68779296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24150, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.10039307624101639, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.012822304712608456, + "grad_norm": 0.07952932268381119, + "kept_token_count": 735.9296875, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9936395525932312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006360440328717232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2986083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.18653259277346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24160, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.10014168173074722, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.011719967192038893, + "grad_norm": 0.10826462507247925, + "kept_token_count": 736.4890625, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9939536273479461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060463806614279745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6278015136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.92954406738284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.09967369139194489, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.012375380098819732, + "grad_norm": 0.10088580846786499, + "kept_token_count": 736.6109375, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9934939861297607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006505997199565172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.6153625488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3782501220703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09983043372631073, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.009888032218441367, + "grad_norm": 0.08509083092212677, + "kept_token_count": 736.64375, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9934745371341706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006525468267500401, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.28291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4648468017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24190, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.09986593648791313, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.012859809212386608, + "grad_norm": 0.10560612380504608, + "kept_token_count": 736.5875, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.993120265007019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006879744306206703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6134429931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5392211914062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24200, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.1004794828593731, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.012179957190528511, + "grad_norm": 0.1306031346321106, + "kept_token_count": 736.275, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933549582958221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006645029969513416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.84178771972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3651489257812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24210, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.10008231997489929, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.01181229748763144, + "grad_norm": 0.1608978509902954, + "kept_token_count": 736.3625, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.015, + "mgd_cos_sim": 0.993573135137558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064268609508872036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0669830322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.63816528320314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24220, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10095047503709793, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.010031648725271226, + "grad_norm": 0.07408994436264038, + "kept_token_count": 735.43125, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9937380850315094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062619071453809735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.03810119628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.31484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24230, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.10012838765978813, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.010293955495581032, + "grad_norm": 0.09952691197395325, + "kept_token_count": 736.4390625, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.993841701745987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006158293411135674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.4663421630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.72540283203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24240, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.09996879771351815, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.011268986575305461, + "grad_norm": 0.0899604856967926, + "kept_token_count": 736.53125, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933264851570129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066735193133354185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.59727783203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.44017028808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24250, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.09976455643773079, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.010611506551504135, + "grad_norm": 0.07689434289932251, + "kept_token_count": 736.5984375, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9939168334007263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060831581242382525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.91163330078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.69928588867185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24260, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.10028037875890732, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.01151019399985671, + "grad_norm": 0.07518414407968521, + "kept_token_count": 736.4515625, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9935249447822571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006475063879042864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1906280517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8741912841797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24270, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.09910218939185142, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.010793370148167013, + "grad_norm": 0.08962948620319366, + "kept_token_count": 737.4125, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9939395964145661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006060414295643568, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.28310241699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.84743347167966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24280, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.09988450855016709, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.013397844880819321, + "grad_norm": 0.10269961506128311, + "kept_token_count": 736.575, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9931774854660034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068225041963160035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.72162170410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.03159484863284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24290, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.10019414573907852, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.011933019338175654, + "grad_norm": 0.09495702385902405, + "kept_token_count": 736.146875, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9940348565578461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005965142976492643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.638525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.6657775878906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24300, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.09980291575193405, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.01305441246367991, + "grad_norm": 0.09746570885181427, + "kept_token_count": 736.6, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9942099392414093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005790068488568068, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.13715209960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2013458251953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24310, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.1003442458808422, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.012350969575345516, + "grad_norm": 0.10118820518255234, + "kept_token_count": 735.9671875, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9937414944171905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258485186845064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.89125671386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.79242248535155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24320, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.09960871264338493, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.010484748380258679, + "grad_norm": 0.09097117185592651, + "kept_token_count": 736.409375, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9938021123409271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00619788458570838, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.553564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.66924438476565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24330, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.10027897655963898, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.011523728119209408, + "grad_norm": 0.08916561305522919, + "kept_token_count": 736.190625, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9942637801170349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005736210756003857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.75905151367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.02464599609374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24340, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.0995689146220684, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.012059647077694535, + "grad_norm": 0.09512315690517426, + "kept_token_count": 736.809375, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9934632480144501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006536734662950039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1222198486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3313903808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24350, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09957160726189614, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.011935824435204268, + "grad_norm": 0.10166247934103012, + "kept_token_count": 736.771875, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9940737009048461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005926294345408678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9386474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5493560791016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10030272528529167, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.011055175913497806, + "grad_norm": 0.08930914103984833, + "kept_token_count": 736.059375, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9936035513877869, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006396440882235765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0980621337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5475769042969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24370, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.10010104700922966, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.010372759494930505, + "grad_norm": 0.08420013636350632, + "kept_token_count": 736.18125, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.993915343284607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00608465513214469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.59091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3323455810547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24380, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.10069135874509812, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.011105565121397376, + "grad_norm": 0.12593266367912292, + "kept_token_count": 736.0265625, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937338769435883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006266120821237564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.30913696289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0693420410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24390, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.1003331147134304, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.01180337304249406, + "grad_norm": 0.14663706719875336, + "kept_token_count": 736.0984375, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9936727166175843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006327282078564167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2872314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.31328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24400, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.10012527629733085, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.013483266159892083, + "grad_norm": 0.12605217099189758, + "kept_token_count": 736.240625, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9937026917934417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006297309696674347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4282531738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4260589599609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24410, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.09989727213978768, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.010940195387229323, + "grad_norm": 0.09989951550960541, + "kept_token_count": 736.265625, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9936910390853881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063089682720601555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1977203369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.28163757324216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24420, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.09958543479442597, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.012966666836291552, + "grad_norm": 0.1089058369398117, + "kept_token_count": 736.9953125, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.993675947189331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006324048899114132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.66949462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1710479736328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24430, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.09958896040916443, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.012253367574885488, + "grad_norm": 0.10728321969509125, + "kept_token_count": 736.5546875, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9937826097011566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006217390112578869, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9510467529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.87037048339846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24440, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.1003171108663082, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.015575964143499732, + "grad_norm": 0.08384622633457184, + "kept_token_count": 736.2953125, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9935711205005646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006428882852196694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3643280029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.77000122070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24450, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.09985478967428207, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.008533350517973303, + "grad_norm": 0.09172742068767548, + "kept_token_count": 736.2125, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9939151883125306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084803864359856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3699127197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8740661621094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24460, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.09985521957278251, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.009377395873889327, + "grad_norm": 0.07242567837238312, + "kept_token_count": 736.4625, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9940517246723175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005948284640908241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.57196044921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5598205566406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24470, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09931970164179801, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.012448561331257223, + "grad_norm": 0.09896479547023773, + "kept_token_count": 737.0828125, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9939713716506958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006028618104755879, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3983184814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.52980041503906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24480, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.09975109323859215, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.009661608329042792, + "grad_norm": 0.07253383845090866, + "kept_token_count": 736.55625, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939387619495392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006061240751296282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6914459228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.63364868164064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.09895769208669662, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.012575280200690032, + "grad_norm": 0.07075551897287369, + "kept_token_count": 737.246875, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.993617582321167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063824311830103396, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7777618408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.98316650390626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24500, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.10027767196297646, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.011749510420486331, + "grad_norm": 0.09333118796348572, + "kept_token_count": 736.3140625, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9939133048057556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006086678057909012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4614990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0035095214844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24510, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.09961875155568123, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.010564787872135639, + "grad_norm": 0.08899501711130142, + "kept_token_count": 736.9859375, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9936619162559509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063380741514265536, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.984619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7999237060547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24520, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.10004211962223053, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.010569800809025764, + "grad_norm": 0.07752127200365067, + "kept_token_count": 736.3046875, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9940228879451751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059771062806248665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6690246582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.87957153320315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.10011607483029365, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.011351233115419745, + "grad_norm": 0.08441608399152756, + "kept_token_count": 736.565625, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9938398718833923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006160133518278599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1893371582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1499969482422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24540, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.09969722926616668, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.01131850159727037, + "grad_norm": 0.06274677067995071, + "kept_token_count": 736.7125, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934797823429108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520221196115017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8582397460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.90769653320314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24550, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09930062070488929, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.011540242424234747, + "grad_norm": 0.10062208771705627, + "kept_token_count": 736.853125, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9935691714286804, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00643082307651639, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3552581787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9825012207031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24560, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.10019632652401925, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.01058232830837369, + "grad_norm": 0.10560568422079086, + "kept_token_count": 736.4359375, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9939200043678283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006079994700849056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.11895141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4024627685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24570, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.09941841661930084, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.011308467807248234, + "grad_norm": 0.08603986352682114, + "kept_token_count": 736.9984375, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936120748519898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006387914065271616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.84727172851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9477142333984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24580, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.10045938193798065, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.011197123862802982, + "grad_norm": 0.13006213307380676, + "kept_token_count": 736.0171875, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933876395225525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612369790673256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2077575683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7160980224609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24590, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.0998289629817009, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.011363501101732254, + "grad_norm": 0.08042817562818527, + "kept_token_count": 736.6640625, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9939360618591309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060639259405434135, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7845520019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1306884765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24600, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.09988391995429993, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.010217848513275386, + "grad_norm": 0.0977712944149971, + "kept_token_count": 736.5828125, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9939416468143463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006058347783982754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.41408386230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5830993652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24610, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10069385319948196, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.01119482098147273, + "grad_norm": 0.09849818050861359, + "kept_token_count": 735.690625, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9937625706195832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237436551600695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.89991760253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9818176269531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24620, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.10038024559617043, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.010569271119311451, + "grad_norm": 0.08178681135177612, + "kept_token_count": 736.0703125, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9939189612865448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081024464219808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.70285949707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1731689453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24630, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.09978862404823304, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.010238054487854243, + "grad_norm": 0.10707245767116547, + "kept_token_count": 736.7203125, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9938340961933136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006165890302509069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.87262268066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.1087188720703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24640, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.10012331828474999, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.009414807241410018, + "grad_norm": 0.07385501265525818, + "kept_token_count": 736.021875, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.993992018699646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006007997784763575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.77520446777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9831085205078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24650, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.09997028112411499, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.012035360466688872, + "grad_norm": 0.0905623808503151, + "kept_token_count": 736.4671875, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9937151551246644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006284844875335693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3341888427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.10960693359374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.10061870813369751, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.010347812995314598, + "grad_norm": 0.10173002630472183, + "kept_token_count": 735.859375, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937534630298615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00624654283747077, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.50152893066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.01405029296876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24670, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.10027147084474564, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.010240454785525798, + "grad_norm": 0.14134536683559418, + "kept_token_count": 736.071875, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9933687031269074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00663127452135086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2280975341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9127624511719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24680, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.10016262829303742, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.012417726498097182, + "grad_norm": 0.11515524238348007, + "kept_token_count": 736.58125, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932310879230499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006768910121172667, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2605804443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7140747070313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24690, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.10041591823101044, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.01089532277546823, + "grad_norm": 0.11186469346284866, + "kept_token_count": 735.984375, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9937316715717316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006268330011516809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1654479980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5555084228516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24700, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.09993057623505593, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.012999028619378804, + "grad_norm": 0.10770559310913086, + "kept_token_count": 736.2953125, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9934363663196564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065636361949145796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5187744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.3854461669922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24710, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.09961176440119743, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010044211847707629, + "grad_norm": 0.07297572493553162, + "kept_token_count": 736.6, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9934317648410798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006568235903978347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.571923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.45778198242186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.09982399716973304, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.01081006322056055, + "grad_norm": 0.10413879156112671, + "kept_token_count": 736.6703125, + "learning_rate": 2.46152806187393e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936907649040222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006309228297322988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.64667358398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.22833251953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24730, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.10037654042243957, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.012178487982600928, + "grad_norm": 0.09655740857124329, + "kept_token_count": 736.4609375, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9937823295593262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006217661593109369, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5037384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.39239807128905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24740, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.1006891280412674, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.011207935679703952, + "grad_norm": 0.11128682643175125, + "kept_token_count": 736.278125, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9930996596813202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006900348421186209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.06732177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.23968505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24750, + "valid_token_count": 818.7125 + }, + { + "actual_token_mask_ratio": 0.09991193264722824, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.010612724535167217, + "grad_norm": 0.19710876047611237, + "kept_token_count": 736.3, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.99354567527771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006454335525631905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7974517822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1190246582031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24760, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.09966003969311714, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.012341246334835887, + "grad_norm": 0.09722093492746353, + "kept_token_count": 736.88125, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9936714112758637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006328595429658889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6647521972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7617614746094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24770, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.10016045123338699, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.010601158812642097, + "grad_norm": 0.0662546455860138, + "kept_token_count": 736.5328125, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9938554763793945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006144523620605469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2449035644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0002807617187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24780, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.10003318265080452, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.012418461777269841, + "grad_norm": 0.09410445392131805, + "kept_token_count": 736.5359375, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933440685272217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006655933894217014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.50885314941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.50303955078124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24790, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.09957391023635864, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.011367261549457908, + "grad_norm": 0.09754418581724167, + "kept_token_count": 736.8859375, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936820805072785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317913252860308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4893463134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.17152099609376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24800, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.10020716935396194, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.011598242772743106, + "grad_norm": 0.12142907083034515, + "kept_token_count": 736.44375, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9938322722911834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006167728453874588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0269744873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1961334228516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24810, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.10042994022369385, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012158402241766453, + "grad_norm": 0.13021615147590637, + "kept_token_count": 736.2796875, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993563038110733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006436951830983162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.09111938476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.64305725097654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24820, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.10046234354376793, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.011677479464560747, + "grad_norm": 0.09945257008075714, + "kept_token_count": 735.940625, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9937424659729004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257517263293266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9497589111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.81719970703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24830, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.09959607943892479, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.01055391514673829, + "grad_norm": 0.08339401334524155, + "kept_token_count": 736.65625, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9937959015369415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204098276793957, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9637939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.64031982421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24840, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.10008036121726036, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.0109229804482311, + "grad_norm": 0.11439963430166245, + "kept_token_count": 736.259375, + "learning_rate": 2.353720475835494e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938570618629455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061429338529706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.34966430664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5415466308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24850, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.09947650581598282, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.016447518579661848, + "grad_norm": 0.11171910911798477, + "kept_token_count": 737.128125, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0198, + "mgd_cos_sim": 0.9932991087436676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006700903829187155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8065124511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6158935546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24860, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.09989161267876626, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.011238233232870699, + "grad_norm": 0.14160215854644775, + "kept_token_count": 736.690625, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9936007916927337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006399213522672653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4491912841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.70196838378905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24870, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.10023901164531708, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.011852592509239913, + "grad_norm": 0.1287337988615036, + "kept_token_count": 736.1875, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9937070310115814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006292964424937963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.48961486816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0061798095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24880, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.10056965351104737, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.011585177434608341, + "grad_norm": 0.10394462943077087, + "kept_token_count": 736.1515625, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933697998523712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006630195118486881, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4586181640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3429016113281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24890, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.10000401064753532, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.011138760391622782, + "grad_norm": 0.09606630355119705, + "kept_token_count": 736.65625, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933375000953675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066625095903873445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.11484680175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.01634216308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24900, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.10096880868077278, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.012723381724208593, + "grad_norm": 0.09075766801834106, + "kept_token_count": 735.778125, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9940290033817292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005970990844070911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1274841308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7422821044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24910, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.10011502578854561, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.012588382931426167, + "grad_norm": 0.12372345477342606, + "kept_token_count": 736.525, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9935403823852539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064596239477396015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.75510864257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4634521484375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24920, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09985804036259652, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.012496190425008536, + "grad_norm": 0.10161573439836502, + "kept_token_count": 736.7359375, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9937212407588959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278772000223398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8617828369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9946990966797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24930, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.10034274756908416, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.011997409723699093, + "grad_norm": 0.12422873079776764, + "kept_token_count": 735.9890625, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9938282549381257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061717437580227855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4329895019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1179626464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24940, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.09926705956459045, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.01012193812057376, + "grad_norm": 0.08768021315336227, + "kept_token_count": 736.959375, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9937825858592987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006217403989285231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8996520996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.07787170410154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24950, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.09986271783709526, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.012235449627041817, + "grad_norm": 0.09107325971126556, + "kept_token_count": 736.515625, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9936581313610077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341868732124567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.44471130371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0974487304687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24960, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09943555369973182, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.011994654033333062, + "grad_norm": 0.09370673447847366, + "kept_token_count": 736.8484375, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9933551967144012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00664480933919549, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.34833068847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7803466796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.09966578856110572, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.01046834266744554, + "grad_norm": 0.08599540591239929, + "kept_token_count": 736.50625, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9941134631633759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005886541586369276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.86309204101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6149536132813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24980, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.0997279018163681, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.010500503703951836, + "grad_norm": 0.07244230806827545, + "kept_token_count": 736.45625, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9938617467880249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00613825311884284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.69407653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.69007263183596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 24990, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.10050876587629318, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.013915325608104467, + "grad_norm": 0.09641974419355392, + "kept_token_count": 736.1328125, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9941784381866455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005821571312844753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6718017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.32862854003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25000, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.10012329444289207, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.010919352946802973, + "grad_norm": 0.10721641033887863, + "kept_token_count": 736.2609375, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935240089893341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006475983466953039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6449951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.61504821777345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25010, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.09968279302120209, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.010351046826690436, + "grad_norm": 0.11753857880830765, + "kept_token_count": 736.665625, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936572968959808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006342707015573978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7920623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2579345703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25020, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09993188828229904, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.012333100009709597, + "grad_norm": 0.07061362266540527, + "kept_token_count": 736.128125, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9931061387062072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006893866136670113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.945947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4016906738281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25030, + "valid_token_count": 817.859375 + }, + { + "actual_token_mask_ratio": 0.10068429186940193, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.013129053125157952, + "grad_norm": 0.0896645188331604, + "kept_token_count": 735.8953125, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9937521278858185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247872114181518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5094940185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7352783203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25040, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.10051155462861061, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.01034656334668398, + "grad_norm": 0.07955963164567947, + "kept_token_count": 735.8421875, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9939204573631286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006079533509910107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3385986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8349365234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25050, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.09972524493932725, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.012868781248107553, + "grad_norm": 0.08402927219867706, + "kept_token_count": 736.4, + "learning_rate": 2.170392717566534e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9937987804412842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006201212201267481, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2514617919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.92625427246094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25060, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.09962610378861428, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.01095415367744863, + "grad_norm": 0.13368749618530273, + "kept_token_count": 736.64375, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9933346927165985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006665305327624082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.51585998535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7393737792969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25070, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.09979967325925827, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.014071501046419143, + "grad_norm": 0.08800757676362991, + "kept_token_count": 736.484375, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9939981818199157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006001811660826206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1491180419922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.0996430054306984, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010719319758936762, + "grad_norm": 0.10327676683664322, + "kept_token_count": 736.80625, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937023639678955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006297623459249735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4077972412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.15480346679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25090, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.09991810247302055, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.010515682119876146, + "grad_norm": 0.12284114211797714, + "kept_token_count": 736.5625, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9936386346817017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006361366063356399, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.02063903808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.13115844726565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25100, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.09986855536699295, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.012630559131503105, + "grad_norm": 0.10636139661073685, + "kept_token_count": 736.5046875, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9937984228134156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062015742063522335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.06423034667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3055145263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25110, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.09988866746425629, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.009943846426904202, + "grad_norm": 0.12876291573047638, + "kept_token_count": 736.4234375, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.993651932477951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006348073855042458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.50035705566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4072998046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25120, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.10004608556628228, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.012665100628510118, + "grad_norm": 0.162820503115654, + "kept_token_count": 736.303125, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931381642818451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068618391640484335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5243865966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.94916381835935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25130, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.09973032400012016, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.012402271945029497, + "grad_norm": 0.08496136963367462, + "kept_token_count": 736.3609375, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9936550736427308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006344924401491881, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.33197021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9529632568359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25140, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.09959921091794968, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.010573601629585028, + "grad_norm": 0.07494591921567917, + "kept_token_count": 736.8671875, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9935603380203247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006439656298607588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1081512451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0810852050781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25150, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.09973872825503349, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.011171163991093636, + "grad_norm": 0.10221373289823532, + "kept_token_count": 737.0421875, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9938436448574066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156343128532171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2726593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.89032287597655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25160, + "valid_token_count": 818.696875 + }, + { + "actual_token_mask_ratio": 0.0997062511742115, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.010499403718858958, + "grad_norm": 0.10205279290676117, + "kept_token_count": 736.421875, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9932859122753144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006714100390672684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2407623291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3621032714844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.10069708898663521, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.011781833972781897, + "grad_norm": 0.14444322884082794, + "kept_token_count": 736.08125, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9938327729701996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006167238019406796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0334503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4875091552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25180, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.10006337612867355, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.011527296714484691, + "grad_norm": 0.09285598993301392, + "kept_token_count": 736.24375, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9940606772899627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005939318705350161, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.50266723632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.61216430664064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25190, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.09983045905828476, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.011358861904591322, + "grad_norm": 0.0704820528626442, + "kept_token_count": 736.73125, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.993555074930191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444919016212225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8748840332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.09697265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25200, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.10047385320067406, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.012080180086195469, + "grad_norm": 0.09254393726587296, + "kept_token_count": 736.00625, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936971426010132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006302844081073999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9808837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8146270751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25210, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.0998007446527481, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.011416619876399636, + "grad_norm": 0.15222640335559845, + "kept_token_count": 736.3796875, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9939810812473298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006018909718841314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0522735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.4584014892578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25220, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.09991355612874031, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.011195874679833651, + "grad_norm": 0.06726084649562836, + "kept_token_count": 736.2484375, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933008432388306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00669916020706296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6095672607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.58372497558594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25230, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.10007260963320733, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.010509637044742703, + "grad_norm": 0.06503424793481827, + "kept_token_count": 736.2375, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9941011250019074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005898898560553789, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5805389404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.26373596191405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25240, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.09989306107163429, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.011987374210730196, + "grad_norm": 0.0924302488565445, + "kept_token_count": 736.5171875, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.993804681301117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006195309106260538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.34291076660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9960479736328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25250, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09971537217497825, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.011374723678454757, + "grad_norm": 0.08814000338315964, + "kept_token_count": 736.4, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9933436572551727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006656345073133707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.21588439941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9871002197266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25260, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.09979984387755395, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.012757428735494614, + "grad_norm": 0.11664613336324692, + "kept_token_count": 736.765625, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937826275825501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062173685058951374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.49774169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7112213134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25270, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.09967126101255416, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.010425386019051075, + "grad_norm": 0.08036714792251587, + "kept_token_count": 736.8953125, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9930753409862518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00692465603351593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7471038818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5182312011719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25280, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.10005040615797042, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.01039080643095076, + "grad_norm": 0.06610233336687088, + "kept_token_count": 735.978125, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9938399493694305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006160045973956585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.76011962890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3888336181641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25290, + "valid_token_count": 817.7984375 + }, + { + "actual_token_mask_ratio": 0.10028137415647506, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.010596211208030582, + "grad_norm": 0.06954772025346756, + "kept_token_count": 736.134375, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9938655138015747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006134464405477047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5899322509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2653472900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.09953021109104157, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.011550945322960615, + "grad_norm": 0.1003103107213974, + "kept_token_count": 737.1453125, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9936526954174042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063473120331764225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.53973388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3258392333984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25310, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.10030746906995773, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.010025976784527301, + "grad_norm": 0.08065184205770493, + "kept_token_count": 736.3859375, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9940938532352448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005906144063919783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9745788574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5952423095703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25320, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.10051410049200057, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.012045505084097385, + "grad_norm": 0.11543820053339005, + "kept_token_count": 736.2125, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9938247263431549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006175263039767742, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6494506835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.64866943359374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25330, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.09999793544411659, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.011629313882440328, + "grad_norm": 0.09779929369688034, + "kept_token_count": 736.0578125, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9936233520507812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006376654282212258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4528289794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5840545654297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25340, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.09974288567900658, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012044763239100576, + "grad_norm": 0.09058275073766708, + "kept_token_count": 736.590625, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9938814401626587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006118562165647745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3003234863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.022900390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09939960017800331, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.010335117438808083, + "grad_norm": 0.06995835155248642, + "kept_token_count": 737.021875, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9934787511825561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006521245371550322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9004638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2307403564453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25360, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.09975804686546326, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.01118738385848701, + "grad_norm": 0.11832809448242188, + "kept_token_count": 736.628125, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9938795983791351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006120402738451958, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4775726318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4719207763672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25370, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09942208454012871, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.011287823179736733, + "grad_norm": 0.10141757875680923, + "kept_token_count": 736.9265625, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935512721538544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448725704103708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.34929809570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8989624023437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25380, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.10000725984573364, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.011519607855007052, + "grad_norm": 0.10764572769403458, + "kept_token_count": 736.31875, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9932245016098022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006775495409965515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3108642578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7329406738281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25390, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.09921938329935073, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.009766306402161718, + "grad_norm": 0.0947626382112503, + "kept_token_count": 737.134375, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937034368515014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006296568457037211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6094024658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4920654296875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25400, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.10025247484445572, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.0123149155639112, + "grad_norm": 0.1005546897649765, + "kept_token_count": 736.2234375, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9934344351291656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006565564312040806, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.88303833007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9621643066406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25410, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.1004582867026329, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.012959025427699089, + "grad_norm": 0.08576637506484985, + "kept_token_count": 736.4046875, + "learning_rate": 1.87216287084645e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9940051853656768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059948178939521314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6803955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3739990234375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25420, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.10015235841274261, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.010779916355386377, + "grad_norm": 0.07707453519105911, + "kept_token_count": 736.38125, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9938624143600464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006137574277818203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5727264404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.92046813964845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25430, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.10013384744524956, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.009635036019608379, + "grad_norm": 0.09449596703052521, + "kept_token_count": 736.2203125, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9934097945690155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006590209994465112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5413330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1249725341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25440, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.0995807908475399, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.0118644745554775, + "grad_norm": 0.08422019332647324, + "kept_token_count": 737.19375, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9935558617115021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444130651652813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0666137695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5724212646484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25450, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.10030273124575614, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.009581166505813598, + "grad_norm": 0.12461137771606445, + "kept_token_count": 736.1921875, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.993847006559372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061530075967311856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.95594177246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8848815917969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25460, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.09968944564461708, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.0120223184581846, + "grad_norm": 0.08916774392127991, + "kept_token_count": 736.659375, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9935954451560974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006404544413089752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.31956481933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.39979858398436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25470, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.10066435709595681, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.01282380148768425, + "grad_norm": 0.08503636717796326, + "kept_token_count": 735.7703125, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9935325920581818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006467406917363405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.73810424804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7715789794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25480, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.09981852620840073, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.011257376847788691, + "grad_norm": 0.07143831998109818, + "kept_token_count": 736.328125, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937352478504181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006264740694314241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4056732177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.95672912597655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25490, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.0999642439186573, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.012475436087697744, + "grad_norm": 0.12952977418899536, + "kept_token_count": 736.6421875, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9936014592647553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063985343091189865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2998840332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.79029541015626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.10030811429023742, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.011160570895299315, + "grad_norm": 0.11333117634057999, + "kept_token_count": 736.2703125, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9939566552639008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006043328065425157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.05933532714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.92333984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25510, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.09977244958281517, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.011263619316741824, + "grad_norm": 0.0868983194231987, + "kept_token_count": 736.90625, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935876607894898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064123374409973625, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3009857177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.772216796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25520, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.10062702521681785, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.012718767346814274, + "grad_norm": 0.09091345965862274, + "kept_token_count": 735.8875, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9936297595500946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006370241474360228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.34918212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.76133422851564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25530, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.09976333454251289, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.011341364355757833, + "grad_norm": 0.19695493578910828, + "kept_token_count": 736.70625, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936252713203431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006374738644808531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7344177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.05284423828124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25540, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.09987065643072128, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012053960561752319, + "grad_norm": 0.08660916984081268, + "kept_token_count": 736.5234375, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.993408203125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006591796316206456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.10508117675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6903411865234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25550, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.09922327622771263, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.012407641997560859, + "grad_norm": 0.12246876209974289, + "kept_token_count": 737.2734375, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9935756027698517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064243866130709645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5732177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.30552978515624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25560, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.10022194907069207, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.012631441932171583, + "grad_norm": 0.10762983560562134, + "kept_token_count": 736.209375, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9936361312866211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006363850273191929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4392028808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9088562011719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25570, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10039799138903618, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.011668582307174801, + "grad_norm": 0.13334870338439941, + "kept_token_count": 736.309375, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9935257375240326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00647425428032875, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6158386230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1842010498047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25580, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.09999893382191657, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.013017903780564666, + "grad_norm": 0.09419621527194977, + "kept_token_count": 736.4625, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9935902833938599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006409725360572338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1686737060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.39405517578126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25590, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.10013339221477509, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.012050420371815562, + "grad_norm": 0.10363475978374481, + "kept_token_count": 736.640625, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936444222927093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006355582177639008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9455871582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.54461364746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25600, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.10032190382480621, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.012256502779200673, + "grad_norm": 0.11297617107629776, + "kept_token_count": 735.975, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9932071983814239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006792809721082449, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.43895568847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.94557495117186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25610, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.09985594376921654, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.01039154827594757, + "grad_norm": 0.10602889955043793, + "kept_token_count": 736.678125, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936852633953095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006314741913229227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2766479492187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2705017089844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.10044335797429085, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.012118735909461975, + "grad_norm": 0.1484411507844925, + "kept_token_count": 736.334375, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9938237428665161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006176262721419334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.59844360351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.36521301269534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25630, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.09871845096349716, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.009941249433904887, + "grad_norm": 0.08338569849729538, + "kept_token_count": 737.234375, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9936908423900604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063091634772717954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.52952880859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5075439453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25640, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.1004429206252098, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.012042966019362212, + "grad_norm": 0.1094772219657898, + "kept_token_count": 736.0546875, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9932890474796295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710972171276808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3326690673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7117462158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25650, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.09986119046807289, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.009976949309930205, + "grad_norm": 0.0919330045580864, + "kept_token_count": 736.64375, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935329616069793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006467028055340052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3285766601563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2794616699219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25660, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.09984081611037254, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.011666913982480764, + "grad_norm": 0.14843499660491943, + "kept_token_count": 736.6265625, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9941325545310974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00586744025349617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.74509887695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0954254150391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25670, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.10094110071659088, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.013051800522953272, + "grad_norm": 0.12051671743392944, + "kept_token_count": 735.425, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9936678588390351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006332151219248771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.55327758789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6745300292969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25680, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.09995069280266762, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.011588732060045005, + "grad_norm": 0.10920458287000656, + "kept_token_count": 736.7390625, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.993412834405899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006587154790759086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1107452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.33660583496095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25690, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.09972196444869041, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.011393972439691424, + "grad_norm": 0.07689289003610611, + "kept_token_count": 736.5875, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9936283469200134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371654942631722, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3459716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.39683532714844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25700, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.10029600709676742, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.011902403412386775, + "grad_norm": 0.09273089468479156, + "kept_token_count": 735.859375, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934765338897705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006523462571203708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4261932373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.84736328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25710, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.10015438050031662, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.010944087756797671, + "grad_norm": 0.10740908980369568, + "kept_token_count": 736.53125, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.993781179189682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006218816339969635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.67889099121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.941259765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25720, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.10013349950313569, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.010392673686146736, + "grad_norm": 0.07751727104187012, + "kept_token_count": 736.2078125, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937868535518646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006213133689016104, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9246551513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9426208496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25730, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.10020406395196915, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.012371125351637601, + "grad_norm": 0.1001695916056633, + "kept_token_count": 736.06875, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9937162041664124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628379201516509, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4045440673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3574920654297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25740, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.1003412202000618, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.010407075146213174, + "grad_norm": 0.08837201446294785, + "kept_token_count": 735.9921875, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9933868885040283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613120529800653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5705841064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.46189575195314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25750, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.09997459650039672, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.010842094849795102, + "grad_norm": 0.1061244085431099, + "kept_token_count": 736.50625, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936886906623841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311309523880482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5826690673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.02525329589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25760, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.10052356198430061, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.011506468476727604, + "grad_norm": 0.0867396891117096, + "kept_token_count": 735.99375, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9936821281909942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00631787246093154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9246368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7416625976563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09985904842615127, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.010785704711452127, + "grad_norm": 0.10239706188440323, + "kept_token_count": 736.475, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9938070178031921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192985735833645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.61509399414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.17966918945314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25780, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.10008401051163673, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.01339247366413474, + "grad_norm": 0.09662635624408722, + "kept_token_count": 736.240625, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9935493707656861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006450624391436577, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.83641052246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2721008300781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25790, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.09979983642697335, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.011104752216488123, + "grad_norm": 0.07812822610139847, + "kept_token_count": 736.378125, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9936393916606903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006360604241490364, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.69147338867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2943542480469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25800, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.0989015482366085, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.010744117433205247, + "grad_norm": 0.09742137044668198, + "kept_token_count": 737.39375, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937878906726837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006212105043232441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.61211853027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.222802734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25810, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.09927924871444702, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.011031056800857187, + "grad_norm": 0.07919727265834808, + "kept_token_count": 736.953125, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937013924121857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006298591941595077, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1725769042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.44775390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25820, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.1001785397529602, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.010600477503612638, + "grad_norm": 0.06293059140443802, + "kept_token_count": 736.4578125, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.014, + "mgd_cos_sim": 0.993271780014038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006728222779929638, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7760009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.25730285644534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25830, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.09984483495354653, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.012189790513366461, + "grad_norm": 0.15110434591770172, + "kept_token_count": 736.6359375, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9935138940811157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006486104428768158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.09222106933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1502319335938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25840, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.1002117283642292, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.010634361393749714, + "grad_norm": 0.08160299062728882, + "kept_token_count": 736.534375, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9930881261825562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006911865063011646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.37635803222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0195770263672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25850, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.09998475164175033, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.012254767771810293, + "grad_norm": 0.084329754114151, + "kept_token_count": 736.4328125, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993752908706665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247105821967125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.124609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6079925537109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25860, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.10010605007410049, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.012309946119785309, + "grad_norm": 0.11640242487192154, + "kept_token_count": 736.7109375, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9932371020317078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006762894801795483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.80357971191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.06329650878905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25870, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.099846600741148, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.010482365591451525, + "grad_norm": 0.11972807347774506, + "kept_token_count": 736.571875, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9932980060577392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006702001020312309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.541064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0786651611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.10055136457085609, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.013144433638080955, + "grad_norm": 0.11440904438495636, + "kept_token_count": 735.703125, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9937402248382569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006259767431765794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2104736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.61517944335935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25890, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.09996603429317474, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.01299711992032826, + "grad_norm": 0.12715014815330505, + "kept_token_count": 736.359375, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9936112523078918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006388738285750151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0673065185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68406066894534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25900, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.1003084622323513, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.010951377591118217, + "grad_norm": 0.08305543661117554, + "kept_token_count": 736.0359375, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938042461872101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006195758376270533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4523132324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5987854003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25910, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.09982287213206291, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.010533819207921624, + "grad_norm": 0.13262739777565002, + "kept_token_count": 736.8390625, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.993653929233551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00634606909006834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2157958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5347869873047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25920, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.09965121895074844, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.010826679784804583, + "grad_norm": 0.09020765870809555, + "kept_token_count": 736.446875, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9938171803951263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006182825192809105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.55101318359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5892059326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25930, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.09941859543323517, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.011519910907372833, + "grad_norm": 0.09204381704330444, + "kept_token_count": 736.721875, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9936918795108796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00630811108276248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4385131835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6335174560547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25940, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.09970340281724929, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.01099926526658237, + "grad_norm": 0.09348056465387344, + "kept_token_count": 736.6625, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935856938362122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006414302345365286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.72899475097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3313385009766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25950, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.10058128833770752, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.010868770768865943, + "grad_norm": 0.11227268725633621, + "kept_token_count": 736.253125, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9936161875724793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006383813172578812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.34295349121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9553253173828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25960, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.09975449442863464, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.01114194062538445, + "grad_norm": 0.08737931400537491, + "kept_token_count": 736.8578125, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937959849834442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204024329781533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.88953857421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6811553955078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25970, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.09989324659109115, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.013694795267656445, + "grad_norm": 0.09144840389490128, + "kept_token_count": 736.696875, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9936992883682251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006300705950707197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0563140869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.06775817871096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25980, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.10030641406774521, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.012930530030280352, + "grad_norm": 0.07954903692007065, + "kept_token_count": 736.403125, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9933638215065003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066361954435706135, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9010864257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.30662841796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 25990, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.09956484884023667, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.010902238963171839, + "grad_norm": 0.09208343178033829, + "kept_token_count": 736.6, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993328207731247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006671778671443462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0075225830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5901702880859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26000, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.10004387348890305, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.012913441751152277, + "grad_norm": 0.11825992166996002, + "kept_token_count": 736.3359375, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9927979409694672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072020674124360085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9277374267578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.10093606263399124, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.011553466040641069, + "grad_norm": 0.24087373912334442, + "kept_token_count": 735.6203125, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933587372303009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066412600688636305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.22352294921876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.43499755859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.09927417039871216, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.010997318802401424, + "grad_norm": 0.11043237894773483, + "kept_token_count": 737.446875, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9931598484516144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006840147078037262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3824798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0505859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26030, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.09946036115288734, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.012682206090539693, + "grad_norm": 0.09664756059646606, + "kept_token_count": 736.528125, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9935952544212341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064047367312014105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7830841064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.75096130371094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26040, + "valid_token_count": 817.8765625 + }, + { + "actual_token_mask_ratio": 0.09998951330780984, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.010487543744966387, + "grad_norm": 0.08458957821130753, + "kept_token_count": 736.4015625, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9935522735118866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006447723880410194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.76796569824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4089874267578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26050, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.10014901012182235, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.013292004261165858, + "grad_norm": 0.1007804349064827, + "kept_token_count": 736.2125, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.993962025642395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037980690598488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2394104003906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26060, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.10005181953310967, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.010431740432977676, + "grad_norm": 0.09740845859050751, + "kept_token_count": 736.7046875, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9940032720565796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005996732600033283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.44309387207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.92853088378905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26070, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.09993122816085816, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.010978806437924505, + "grad_norm": 0.07381171733140945, + "kept_token_count": 736.13125, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9937175512313843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062824515625834465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6535705566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2155334472656, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26080, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.10005161911249161, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.009544940385967492, + "grad_norm": 0.10128669440746307, + "kept_token_count": 736.4734375, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9936334788799286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366513762623071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.28626403808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.87827453613284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26090, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10014150515198708, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.01115604555234313, + "grad_norm": 0.09640497714281082, + "kept_token_count": 736.5421875, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9936230778694153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006376926880329847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.70264892578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.03570861816405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26100, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.1008930966258049, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.013627913547679782, + "grad_norm": 0.07968417555093765, + "kept_token_count": 735.7796875, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9936884641647339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311525218188762, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3679595947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5257598876953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26110, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.10005702152848243, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.012665781658142805, + "grad_norm": 0.15754885971546173, + "kept_token_count": 736.353125, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9934056043624878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006594391353428364, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6139404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8720672607422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26120, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10031256750226021, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.01090420582331717, + "grad_norm": 0.09767462313175201, + "kept_token_count": 736.053125, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9936412632465362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006358741782605648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0190948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.47966918945315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26130, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.09962523579597474, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.011122703738510608, + "grad_norm": 0.11031713336706161, + "kept_token_count": 736.6640625, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9935145437717438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006485469453036785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.58661499023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.66468200683596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.10012206882238388, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.011769979540258647, + "grad_norm": 0.10470518469810486, + "kept_token_count": 736.321875, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933756113052368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006624392047524452, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2559814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2963165283203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26150, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.10040657594799995, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.011257340153679251, + "grad_norm": 0.09492803364992142, + "kept_token_count": 735.775, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931322693824768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006867739371955395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7848724365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.231591796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26160, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.09963272511959076, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.011521679349243641, + "grad_norm": 0.11259196698665619, + "kept_token_count": 737.0765625, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9939019620418549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006098020169883966, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6209014892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1733001708984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26170, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.10013906881213189, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.010199683764949442, + "grad_norm": 0.08618659526109695, + "kept_token_count": 736.0671875, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9935742139816284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006425792165100574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0220031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5189208984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26180, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.0992660827934742, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.011276447819545865, + "grad_norm": 0.11755121499300003, + "kept_token_count": 736.6109375, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9934532880783081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006546713691204786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.38706359863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2445861816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26190, + "valid_token_count": 817.7875 + }, + { + "actual_token_mask_ratio": 0.10015476495027542, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.011401527095586061, + "grad_norm": 0.16332164406776428, + "kept_token_count": 736.528125, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.993456882238388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006543106306344271, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.36993408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7326416015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26200, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.10033328458666801, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.010354840289801359, + "grad_norm": 0.09115784615278244, + "kept_token_count": 736.275, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936716556549072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006328322738409042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.72962341308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.60765075683594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26210, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.09956941828131675, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.011794246640056372, + "grad_norm": 0.10084322094917297, + "kept_token_count": 736.890625, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931961357593536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00680386545136571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9695129394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.63641357421875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26220, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.10043124780058861, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.011951443646103144, + "grad_norm": 0.08069773018360138, + "kept_token_count": 736.0296875, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.993517404794693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006482600700110197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.95960388183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4735809326172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.1003089651465416, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.011096991412341595, + "grad_norm": 0.10412067174911499, + "kept_token_count": 735.965625, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.994022011756897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005977972410619259, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9946624755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.73305053710936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26240, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.10007703080773353, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.009750115359202028, + "grad_norm": 0.09742390364408493, + "kept_token_count": 736.40625, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937420070171357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257984694093466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6614105224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3317047119141, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26250, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.10029681995511056, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.014259153418242932, + "grad_norm": 0.0849933996796608, + "kept_token_count": 736.190625, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.993666136264801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006333871558308601, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5113098144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.74681091308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26260, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.10093124806880951, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.010243372712284326, + "grad_norm": 0.07658801972866058, + "kept_token_count": 735.7421875, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935710549354553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006428937148302794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2613861083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.18218994140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.09961536750197411, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.011865259427577258, + "grad_norm": 0.08941024541854858, + "kept_token_count": 736.8265625, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933145046234131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006685504596680403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9107971191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6852630615234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26280, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.10025302395224571, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.010867299977689981, + "grad_norm": 0.09575698524713516, + "kept_token_count": 735.8328125, + "learning_rate": 1.237623657682847e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938077390193939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192261446267366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.09083557128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.0587982177734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26290, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.0994463786482811, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.009616940980777144, + "grad_norm": 0.08713740855455399, + "kept_token_count": 737.346875, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9936250865459442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006374909821897745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.82774047851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.3064300537109, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26300, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.09981513991951943, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.013951617805287243, + "grad_norm": 0.09220852702856064, + "kept_token_count": 736.8859375, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9939276039600372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006072396133095026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9383514404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.79453430175784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26310, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.09981532096862793, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.010558843147009611, + "grad_norm": 0.09637818485498428, + "kept_token_count": 736.5921875, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9933468997478485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653107423335314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5022857666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.84808044433595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26320, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.10009064301848411, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.012349958764389157, + "grad_norm": 0.06631575524806976, + "kept_token_count": 736.9140625, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.994034743309021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005965251009911299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.150732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.8590393066406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26330, + "valid_token_count": 818.8765625 + }, + { + "actual_token_mask_ratio": 0.09975727126002312, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.01500228838995099, + "grad_norm": 0.10330630093812943, + "kept_token_count": 736.6671875, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9935976386070251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064023657701909546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2361999511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7657775878906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26340, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.10035981759428977, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.0113909550011158, + "grad_norm": 0.09471606463193893, + "kept_token_count": 735.89375, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937415540218353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258439458906651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.50441589355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2319793701172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26350, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.09962888583540916, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.009417903842404485, + "grad_norm": 0.09673063457012177, + "kept_token_count": 736.6234375, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937132775783539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006286739278584719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1096252441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.92088623046874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26360, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.10002045035362243, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.01159224109724164, + "grad_norm": 0.10440713167190552, + "kept_token_count": 736.5421875, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.993930971622467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006069028470665217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4247772216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8110321044922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26370, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.09990908056497574, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.012516156956553458, + "grad_norm": 0.11618462949991226, + "kept_token_count": 736.7140625, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9933811068534851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00661887526512146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.32891845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3378570556641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26380, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.09982831701636315, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.013499885518103837, + "grad_norm": 0.10972320288419724, + "kept_token_count": 736.2890625, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9939241886138916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006075806729495525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8864501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.657568359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26390, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.10054458305239677, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.012492981739342213, + "grad_norm": 0.08998457342386246, + "kept_token_count": 736.1453125, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9935005605220795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006499442365020513, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.46885375976564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26400, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.09971165806055068, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.00951479971408844, + "grad_norm": 0.08285806328058243, + "kept_token_count": 736.9609375, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9935346603393554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006465353351086378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2788787841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.91608276367185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.10042957440018654, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.0102711436804384, + "grad_norm": 0.09680374711751938, + "kept_token_count": 736.0921875, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9934329390525818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006567053217440844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8715423583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.20946350097654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26420, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10008308738470077, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.012293837079778314, + "grad_norm": 0.10446591675281525, + "kept_token_count": 736.265625, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9934519112110138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00654808534309268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.84230041503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.38364868164064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26430, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.1008618265390396, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.011342494701966643, + "grad_norm": 0.08613373339176178, + "kept_token_count": 735.9609375, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9940640449523925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00593596650287509, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1176361083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2592071533203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26440, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.09905546307563781, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.00990790342912078, + "grad_norm": 0.10279715061187744, + "kept_token_count": 737.2796875, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9941994607448578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005800544843077659, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.549560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8944030761719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26450, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.09987871497869491, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.009962078696116806, + "grad_norm": 0.07353442162275314, + "kept_token_count": 736.6046875, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9936679482460022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063320380635559555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.77219848632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1822845458984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26460, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.10046084225177765, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.011906375735998153, + "grad_norm": 0.12347938120365143, + "kept_token_count": 736.0453125, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9935434401035309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006456551421433687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.19854736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1396453857422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26470, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.09892175942659379, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.011778810992836952, + "grad_norm": 0.11224177479743958, + "kept_token_count": 737.2, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.993807178735733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192817725241185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9895263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.94310607910154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26480, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.10045767053961754, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.011040863627567888, + "grad_norm": 0.10490395873785019, + "kept_token_count": 736.15625, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9938247263431549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006175261177122593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5905731201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.90333557128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26490, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.0996307797729969, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.01142190657556057, + "grad_norm": 0.09164419025182724, + "kept_token_count": 736.5734375, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9937255501747131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006274449173361063, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6027862548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.37453308105466, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26500, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.0999126061797142, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.012312217662110925, + "grad_norm": 0.09050117433071136, + "kept_token_count": 736.63125, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9937548637390137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006245127040892839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.806689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1659790039063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26510, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.1004326030611992, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.009809404332190751, + "grad_norm": 0.0975789874792099, + "kept_token_count": 736.3375, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9941407442092896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058592455461621284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1165832519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.09685668945315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26520, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.09991016387939453, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.012310782307758927, + "grad_norm": 0.08060945570468903, + "kept_token_count": 736.2453125, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9941063165664673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005893683340400457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.71018676757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4223999023437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26530, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.0992475226521492, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.01083979750983417, + "grad_norm": 0.07844161242246628, + "kept_token_count": 736.90625, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934723556041718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006527632474899292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.56324462890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0747344970703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26540, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.10047688037157058, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.01134668281301856, + "grad_norm": 0.10299905389547348, + "kept_token_count": 736.1171875, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9938172936439514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061827079392969605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4375366210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7203704833984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26550, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.1006631076335907, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.011214186204597354, + "grad_norm": 0.11570320278406143, + "kept_token_count": 735.9265625, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.993517154455185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006482846103608608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7985137939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8827880859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26560, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.100089131295681, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.015397749375551939, + "grad_norm": 0.09782280772924423, + "kept_token_count": 736.2953125, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0185, + "mgd_cos_sim": 0.9938199937343597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180001422762871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.97539672851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.30104064941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26570, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.10040311962366104, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.01175976349040866, + "grad_norm": 0.10176653414964676, + "kept_token_count": 736.0671875, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9939953982830048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006004600040614605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.03406982421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.46490478515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26580, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.09937109127640724, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.010393617022782565, + "grad_norm": 0.0742873102426529, + "kept_token_count": 736.8640625, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937806189060211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062193660996854305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2400634765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.6263000488281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26590, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.099628047645092, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.013792583486065269, + "grad_norm": 0.13776087760925293, + "kept_token_count": 736.6609375, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9930781960487366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006921812705695629, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.45526428222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.480810546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26600, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.09960205852985382, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.011641515558585525, + "grad_norm": 0.09474077075719833, + "kept_token_count": 736.7, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9938141226768493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061858841218054295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.51233825683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9408721923828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26610, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.0997176043689251, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.009851910173892975, + "grad_norm": 0.11551668494939804, + "kept_token_count": 736.6296875, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9935664296150207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433573085814714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7901611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1591339111328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26620, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10010014474391937, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.012058360502123833, + "grad_norm": 0.09255398064851761, + "kept_token_count": 736.3234375, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9937147498130798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628525922074914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6667724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.42511291503905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26630, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.10051505342125892, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.0120786527171731, + "grad_norm": 0.06631971150636673, + "kept_token_count": 736.2140625, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9939624071121216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037573516368866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2231475830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6299163818359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26640, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.10036424323916435, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.010792776197195052, + "grad_norm": 0.11709890514612198, + "kept_token_count": 736.159375, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9935126900672913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006487311143428087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7504821777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0983856201172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.09974707439541816, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.011129660066217184, + "grad_norm": 0.06924083083868027, + "kept_token_count": 736.5796875, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9937249600887299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00627504251897335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4013305664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3724700927734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26660, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.10041483342647553, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.01100563695654273, + "grad_norm": 0.07120858877897263, + "kept_token_count": 736.140625, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9935574293136596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006442568451166153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.67296142578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.92828979492185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26670, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.10043510943651199, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.010892881732434034, + "grad_norm": 0.10069135576486588, + "kept_token_count": 736.0890625, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9939298868179322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060701116919517515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.39723815917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.32879943847655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26680, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.09959627017378807, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.011304037179797888, + "grad_norm": 0.08986317366361618, + "kept_token_count": 736.940625, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.993655675649643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006344323419034481, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9462188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.50795288085936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26690, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.09910704717040061, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.012028019689023495, + "grad_norm": 0.08056368678808212, + "kept_token_count": 737.3171875, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9937919974327087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006208000145852566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.87966003417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0843841552734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26700, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.09984600096940995, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.011870195996016264, + "grad_norm": 0.08364343643188477, + "kept_token_count": 736.7609375, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9939193189144134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006080686673521995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3709228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7901641845703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26710, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.10025496408343315, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.012454786524176597, + "grad_norm": 0.08442921936511993, + "kept_token_count": 736.075, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0158, + "mgd_cos_sim": 0.9933729648590088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006627029925584793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9329437255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.13439025878904, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.10049471780657768, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.014860103279352188, + "grad_norm": 0.11214551329612732, + "kept_token_count": 735.9890625, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0181, + "mgd_cos_sim": 0.9935615956783295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006438413262367248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.72840576171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6650695800781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26730, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10024515315890312, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.01161734787747264, + "grad_norm": 0.07737595587968826, + "kept_token_count": 735.9921875, + "learning_rate": 9.587191621316154e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9931766211986541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006823381315916777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.256787109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9180572509766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26740, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.10024025961756707, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.010021973168477416, + "grad_norm": 0.06616155803203583, + "kept_token_count": 736.45, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9936884999275207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311512086540461, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2997619628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4871337890625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26750, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.09963494315743446, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.012614594912156463, + "grad_norm": 0.09586974233388901, + "kept_token_count": 736.915625, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0158, + "mgd_cos_sim": 0.993573147058487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006426850333809853, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.482958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2295349121094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26760, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09981002658605576, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.011608017003163695, + "grad_norm": 0.06308889389038086, + "kept_token_count": 736.4703125, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9941185176372528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058814890682697294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.25978393554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.51019287109375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26770, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.09998227804899215, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.010687882453203202, + "grad_norm": 0.06981949508190155, + "kept_token_count": 736.64375, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9937053859233856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062946190126240255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6654815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.01792297363284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26780, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.09995756894350052, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.013356849364936352, + "grad_norm": 0.11719513684511185, + "kept_token_count": 736.5234375, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0166, + "mgd_cos_sim": 0.9936015784740448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006398418266326189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0880889892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9648712158203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.09979436919093132, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.011828282382339239, + "grad_norm": 0.0985509604215622, + "kept_token_count": 736.5703125, + "learning_rate": 9.24127773842518e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9937269568443299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006273040361702442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0617370605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.32691650390626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26800, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.09933153539896011, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.010231827618554235, + "grad_norm": 0.10134124755859375, + "kept_token_count": 737.00625, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9937492668628692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006250738073140383, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7667572021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.34624938964845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.09911723732948304, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.010429387260228396, + "grad_norm": 0.11388753354549408, + "kept_token_count": 737.08125, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9935394525527954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006460561323910952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8635650634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0537506103516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26820, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.10010931193828583, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.011872365465387702, + "grad_norm": 0.05859757959842682, + "kept_token_count": 736.5359375, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9935795426368713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006420450657606125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4821014404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.0280395507813, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26830, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.09986614659428597, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.009803579747676849, + "grad_norm": 0.08452026546001434, + "kept_token_count": 736.5453125, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0128, + "mgd_cos_sim": 0.9940285384654999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005971468612551689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.56419067382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8380584716797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26840, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.10030964761972427, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.011870058998465539, + "grad_norm": 0.10549256205558777, + "kept_token_count": 736.2578125, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9931998074054718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006800199020653963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.74938049316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3474182128906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26850, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09934597685933114, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.011595293553546071, + "grad_norm": 0.10626808553934097, + "kept_token_count": 736.9375, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9938933908939361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006106616556644439, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.29095153808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3999969482422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10062679499387742, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.010526698641479015, + "grad_norm": 0.08491187542676926, + "kept_token_count": 735.8484375, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9936912953853607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006308710109442473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.47984924316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.58180847167966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26870, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.1001114197075367, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.010701752128079533, + "grad_norm": 0.10078414529561996, + "kept_token_count": 736.61875, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9939622700214386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037726253271103, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.44443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.44227905273436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26880, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.09949222430586815, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.010516052087768912, + "grad_norm": 0.07329076528549194, + "kept_token_count": 736.7203125, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9936763644218445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006323636136949062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0750732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3972534179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26890, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.10028635635972023, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.010683323862031103, + "grad_norm": 0.07302885502576828, + "kept_token_count": 736.209375, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9938612759113312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061387218534946445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7995574951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8855773925781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26900, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.10080787688493728, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.010278230626136064, + "grad_norm": 0.10064797103404999, + "kept_token_count": 735.8515625, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9934052348136901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065947665832936765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.83985290527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2133361816406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26910, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.1003830723464489, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.009831296326592565, + "grad_norm": 0.11109083890914917, + "kept_token_count": 736.0875, + "learning_rate": 8.567936562219547e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9937262237071991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00627376614138484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.03111877441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7806671142578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26920, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.1000664360821247, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.011548339901492, + "grad_norm": 0.10703947395086288, + "kept_token_count": 736.3484375, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9938854336738586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006114561203867197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6136016845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1352966308594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26930, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.10040649399161339, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.009301506914198399, + "grad_norm": 0.07894038408994675, + "kept_token_count": 736.1328125, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9940211772918701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059788313694298266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.602685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.09569091796874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26940, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.10034217834472656, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.011250671045854688, + "grad_norm": 0.08289596438407898, + "kept_token_count": 736.1125, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9941112518310546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005888739041984082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6604827880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0070434570313, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26950, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.09991741925477982, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.011894332757219673, + "grad_norm": 0.147968590259552, + "kept_token_count": 736.415625, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9936851799488068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006314810551702976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.60597534179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.9363708496094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26960, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.10008290484547615, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.00878337793983519, + "grad_norm": 0.1520702987909317, + "kept_token_count": 736.2015625, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.994166374206543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005833627376705408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.85328063964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.5593292236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26970, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.09931624308228493, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.011981272464618086, + "grad_norm": 0.07889271527528763, + "kept_token_count": 737.0484375, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9937147676944733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006285255122929811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.66489562988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1048828125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26980, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.10055375471711159, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.011424910183995962, + "grad_norm": 0.10842286050319672, + "kept_token_count": 735.828125, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9935966432094574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006403351202607155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.10597534179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1684326171875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 26990, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.10003268420696258, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.01190600572153926, + "grad_norm": 0.10014324635267258, + "kept_token_count": 736.5578125, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9935714483261109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006428539473563433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6099029541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7086608886719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27000, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.0996388852596283, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.010485579213127493, + "grad_norm": 0.0915072038769722, + "kept_token_count": 736.434375, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9942829728126525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005717010516673326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2417755126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0441192626953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27010, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.09935291707515717, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.011497695278376342, + "grad_norm": 0.06407861411571503, + "kept_token_count": 736.9578125, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9940509259700775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005949071422219276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.910546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8107116699219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27020, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.0997111015021801, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.011899286974221468, + "grad_norm": 0.08772241324186325, + "kept_token_count": 736.7625, + "learning_rate": 7.972443479549035e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9938537240028381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006146277207881213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.45233459472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.29407958984376, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27030, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10012046918272972, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.011899368185549975, + "grad_norm": 0.08290717005729675, + "kept_token_count": 736.60625, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9936903297901154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006309675984084606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.01731567382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6899139404297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27040, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.10070979744195938, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.011362542072311044, + "grad_norm": 0.09251236915588379, + "kept_token_count": 736.1515625, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9937309741973877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006269018817692995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6174713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.73862915039064, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27050, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.09987681284546852, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.011225901544094086, + "grad_norm": 0.08849430084228516, + "kept_token_count": 736.840625, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.993947833776474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006052166782319545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7756286621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0750305175781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27060, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.09936674311757088, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.013358244858682155, + "grad_norm": 0.10642911493778229, + "kept_token_count": 736.9390625, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0168, + "mgd_cos_sim": 0.9931270182132721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006872996781021356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4693237304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0524169921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27070, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.10029633566737176, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.013288151565939188, + "grad_norm": 0.0984724760055542, + "kept_token_count": 736.2328125, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0164, + "mgd_cos_sim": 0.993785846233368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006214152462780476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4680114746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.7876495361328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27080, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.10020083263516426, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.010129431262612343, + "grad_norm": 0.09547316282987595, + "kept_token_count": 736.2953125, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9937246859073638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006275328807532787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.83815612792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.46099548339845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.09985060170292855, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.012250737566500903, + "grad_norm": 0.10012958198785782, + "kept_token_count": 736.3875, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9934013485908508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006598646473139525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.43829650878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9036041259766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27100, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.09964971244335175, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.010299073345959187, + "grad_norm": 0.10598017275333405, + "kept_token_count": 736.784375, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9937606573104858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006239340454339981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.246435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.97832641601565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09985206201672554, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.011074708681553603, + "grad_norm": 0.05887405574321747, + "kept_token_count": 736.5484375, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9940027594566345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005997234955430031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.55541076660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1283325195312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27120, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.09974915906786919, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.011862317426130175, + "grad_norm": 0.1127876415848732, + "kept_token_count": 737.0375, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.993949168920517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006050827261060476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.43207397460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0666442871094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27130, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.09942727908492088, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.010978613840416073, + "grad_norm": 0.09826287627220154, + "kept_token_count": 736.95, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.993470698595047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006529298610985279, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.61261291503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.48395080566405, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27140, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.10053521692752838, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.010276429681107402, + "grad_norm": 0.08715513348579407, + "kept_token_count": 735.834375, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9936312913894654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006368713732808828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.02056579589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4072662353516, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27150, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.1003314882516861, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.009417231241241098, + "grad_norm": 0.09913203120231628, + "kept_token_count": 736.271875, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9937048554420471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006295155640691519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.03701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.27672119140624, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27160, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.10056868642568588, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.010498798824846745, + "grad_norm": 0.08393112570047379, + "kept_token_count": 736.178125, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9940284848213196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005971525888890028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.74584045410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.11663513183595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27170, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.099528269469738, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.013268774561583995, + "grad_norm": 0.1461816281080246, + "kept_token_count": 736.684375, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0165, + "mgd_cos_sim": 0.9936106383800507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006389363203197718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3330108642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.82872314453124, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27180, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.1000646933913231, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.01116193225607276, + "grad_norm": 0.09519384801387787, + "kept_token_count": 736.5109375, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9935115694999694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064884286373853685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0235168457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9731719970703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27190, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.10050413012504578, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.011910280771553516, + "grad_norm": 0.08774877339601517, + "kept_token_count": 736.2140625, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9935962140560151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006403771508485079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.25762634277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8855010986328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27200, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.09977126643061637, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.011962341237813234, + "grad_norm": 0.10928318649530411, + "kept_token_count": 736.825, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.993533480167389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006466515362262726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.48451843261716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.19420471191404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27210, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.09979629963636398, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.011036016838625073, + "grad_norm": 0.10996544361114502, + "kept_token_count": 736.603125, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9937646210193634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062353678047657015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8859130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.90924072265625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09970843568444251, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.010275420546531678, + "grad_norm": 0.10970927029848099, + "kept_token_count": 736.7640625, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.993691885471344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006308103818446398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9143035888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.1238647460938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27230, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.09954487979412079, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.011063513392582536, + "grad_norm": 0.09726283699274063, + "kept_token_count": 736.7984375, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9938447117805481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006155290547758341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.842041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6374877929687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27240, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.0999380111694336, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.011692310543730855, + "grad_norm": 0.08928150683641434, + "kept_token_count": 736.4234375, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9939056396484375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006094367057085037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0844757080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8318603515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.10011175498366356, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.012708351714536548, + "grad_norm": 0.09842496365308762, + "kept_token_count": 736.3, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.993518590927124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006481409911066294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.72907104492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.63042602539065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27260, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10025199726223946, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.010171596752479672, + "grad_norm": 0.12112509459257126, + "kept_token_count": 736.1953125, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9940096974372864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005990302190184593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.99765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.966552734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27270, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.10008844956755639, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.011429841071367264, + "grad_norm": 0.07885988056659698, + "kept_token_count": 736.471875, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9935719788074493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006428018864244223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.22472229003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3881866455078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27280, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.1006995014846325, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.01073993849568069, + "grad_norm": 0.10998602956533432, + "kept_token_count": 735.5734375, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9936831712722778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063168155960738655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.33063049316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.67356567382814, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27290, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.10012543350458145, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.010992382606491446, + "grad_norm": 0.09152640402317047, + "kept_token_count": 736.3078125, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.993654727935791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063452782109379765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.73479919433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4908813476562, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27300, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09973189160227776, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.011080827424302697, + "grad_norm": 0.06949624419212341, + "kept_token_count": 736.7453125, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9938426971435547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006157310865819454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.55601501464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.29278564453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27310, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10003501996397972, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.011430519493296742, + "grad_norm": 0.09507617354393005, + "kept_token_count": 736.565625, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9942111730575561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00578883970156312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.59677734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.08880920410155, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27320, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.09990830868482589, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.011014291876927018, + "grad_norm": 0.09754075109958649, + "kept_token_count": 736.31875, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9941691994667053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005830812640488148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1232940673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.31662902832034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27330, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.10055207833647728, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.01158053739927709, + "grad_norm": 0.057121641933918, + "kept_token_count": 735.9484375, + "learning_rate": 6.406855296131387e-07, + "loss": 0.015, + "mgd_cos_sim": 0.993169641494751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006830361485481262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1638488769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.41566772460936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27340, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.10019844025373459, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.01121084182523191, + "grad_norm": 0.0865095853805542, + "kept_token_count": 736.2359375, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9939276695251464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006072335224598646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3987701416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3336242675781, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27350, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.09949080497026444, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.010399340372532605, + "grad_norm": 0.07757539302110672, + "kept_token_count": 737.1109375, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9938314914703369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006168504059314728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.41588439941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.60296630859375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27360, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.09981262609362603, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.00978493490256369, + "grad_norm": 0.07273528724908829, + "kept_token_count": 736.590625, + "learning_rate": 6.264216762935187e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9936401307582855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006359856575727463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6476654052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.11484680175784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27370, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09998432248830795, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.010754996491596103, + "grad_norm": 0.11746492981910706, + "kept_token_count": 736.3, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.993405568599701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006594420038163662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1937530517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5340118408203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27380, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.10009726583957672, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.00925500518642366, + "grad_norm": 0.05640057474374771, + "kept_token_count": 736.25, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9934341013431549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065658831968903545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2871826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.72059020996096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27390, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.10081518292427064, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.011331043764948845, + "grad_norm": 0.1007353663444519, + "kept_token_count": 735.7734375, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9937979280948639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006202057655900717, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4436309814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9653350830078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.10086055248975753, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.011805367469787598, + "grad_norm": 0.09832819551229477, + "kept_token_count": 735.6078125, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9932064354419708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006793561764061451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.99851684570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3853302001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27410, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.10081925168633461, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.009914559358730913, + "grad_norm": 0.10496534407138824, + "kept_token_count": 735.36875, + "learning_rate": 6.029979684572262e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9937862992286682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006213689874857664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5728240966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.45953979492185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27420, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.10024690479040146, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.011430589994415642, + "grad_norm": 0.09107911586761475, + "kept_token_count": 736.1359375, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9936644196510315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006335571687668562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7253051757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8413482666016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27430, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.10000780150294304, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.012068746844306588, + "grad_norm": 0.11222631484270096, + "kept_token_count": 736.40625, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9939454376697541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006054573226720095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5261474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4997222900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27440, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.09964059367775917, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.012749852286651731, + "grad_norm": 0.09618711471557617, + "kept_token_count": 736.7578125, + "learning_rate": 5.89153622838805e-07, + "loss": 0.016, + "mgd_cos_sim": 0.9935892403125763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006410762015730142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3724365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.34595947265626, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27450, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.1003748320043087, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.012470389157533646, + "grad_norm": 0.10806538164615631, + "kept_token_count": 736.2125, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.99351606965065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064839361235499385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.50487365722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.84227294921874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27460, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.10006452873349189, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.009880156721919775, + "grad_norm": 0.06837093830108643, + "kept_token_count": 736.3046875, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9938784301280975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006121574807912112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1816345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2499542236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27470, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.10036987140774727, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.011132462788373232, + "grad_norm": 0.12492457777261734, + "kept_token_count": 736.2734375, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9938369154930115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006163097824901343, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0984313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2688323974609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27480, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.1000096395611763, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.01001078262925148, + "grad_norm": 0.04912932962179184, + "kept_token_count": 736.203125, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9945609271526337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005439061857759953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1217041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.47641906738284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27490, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.10021347552537918, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.010005503566935659, + "grad_norm": 0.0790497213602066, + "kept_token_count": 736.3921875, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9938406586647034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159338448196649, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.90106506347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5181640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27500, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.0999225564301014, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.010928976535797118, + "grad_norm": 0.11251508444547653, + "kept_token_count": 736.2125, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9934128761291504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006587129831314087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1294860839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.43702087402346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27510, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.10047388374805451, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.012879128102213144, + "grad_norm": 0.09873709827661514, + "kept_token_count": 736.165625, + "learning_rate": 5.574632764857974e-07, + "loss": 0.016, + "mgd_cos_sim": 0.9936691105365754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063308804295957085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.92555541992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.9134460449219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27520, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10003960654139518, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010011664452031254, + "grad_norm": 0.12492728978395462, + "kept_token_count": 736.6375, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9936757683753967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006324241124093533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6601104736328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27530, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.10030206367373466, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.011537664895877242, + "grad_norm": 0.09117017686367035, + "kept_token_count": 736.475, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9937624573707581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237549521028995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4281402587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5528198242188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27540, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.09936687275767327, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.010548799624666571, + "grad_norm": 0.06259511411190033, + "kept_token_count": 736.9375, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9936589956283569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006340995244681835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9792205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0542358398437, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27550, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.10021019876003265, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.013015206251293421, + "grad_norm": 0.09606850892305374, + "kept_token_count": 736.225, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0161, + "mgd_cos_sim": 0.9938079237937927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192081607878208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3293853759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.0844421386719, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27560, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.0989574871957302, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.011445971811190248, + "grad_norm": 0.11775602400302887, + "kept_token_count": 737.5625, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9942374110221863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005762590002268553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.287646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.38116760253905, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27570, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.10034813582897187, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.009191800467669963, + "grad_norm": 0.08672605454921722, + "kept_token_count": 736.3578125, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9937934219837189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006206576246768236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0838104248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4631805419922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27580, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.10034580379724503, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.010297368047758937, + "grad_norm": 0.06390329450368881, + "kept_token_count": 736.0265625, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9937018692493439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006298131309449673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.24949035644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3364288330078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27590, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.09974911287426949, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.011350712785497308, + "grad_norm": 0.10288726538419724, + "kept_token_count": 736.875, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9939155399799346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084440927952528, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1537261962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.20972595214846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27600, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.10000943019986153, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.009495054744184018, + "grad_norm": 0.0831516906619072, + "kept_token_count": 736.3734375, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9939405500888825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006059457082301378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8281768798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3158477783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27610, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.10011672228574753, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.012177834799513221, + "grad_norm": 0.09873443841934204, + "kept_token_count": 736.6515625, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9938299179077148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061700775288045405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8878936767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8672637939453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27620, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.10020779371261597, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.011967665748670697, + "grad_norm": 0.1590590924024582, + "kept_token_count": 736.46875, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9934512972831726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006548700015991926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.24900817871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.101953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27630, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.09944324642419815, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.012749270303174853, + "grad_norm": 0.08698888123035431, + "kept_token_count": 736.9875, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9937335848808289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006266426760703325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3291809082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.7017791748047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27640, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.09972439408302307, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.011149723362177611, + "grad_norm": 0.09525030106306076, + "kept_token_count": 736.6421875, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.993776935338974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006223065312951803, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1357177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.43406066894534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27650, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.09975958168506623, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.011339208390563727, + "grad_norm": 0.10212632268667221, + "kept_token_count": 736.88125, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9934407830238342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006559233926236629, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1448455810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.81727905273436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27660, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.10018147081136704, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.010482730437070131, + "grad_norm": 0.09184592217206955, + "kept_token_count": 736.190625, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9935620367527008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006437967158854007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.32053527832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.109033203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27670, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.09941971078515052, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.011246853368356824, + "grad_norm": 0.0946742370724678, + "kept_token_count": 736.7328125, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9940288424491882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00597114022821188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8599151611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0975372314453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27680, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.0996948540210724, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.01358876507729292, + "grad_norm": 0.08310434222221375, + "kept_token_count": 736.840625, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0169, + "mgd_cos_sim": 0.9934478461742401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552158575505018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2693786621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.15941162109374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27690, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.10015813037753105, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.009625819744542241, + "grad_norm": 0.06980623304843903, + "kept_token_count": 735.9515625, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9939504086971283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006049589160829783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6522644042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7366607666016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27700, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.09987308606505393, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.012350235972553492, + "grad_norm": 0.11666234582662582, + "kept_token_count": 736.2859375, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9935664296150207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433576624840498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.545654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.96705932617186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27710, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.09929331690073014, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.014689097460359336, + "grad_norm": 0.10400750488042831, + "kept_token_count": 737.065625, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0179, + "mgd_cos_sim": 0.9935771942138671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006422812957316637, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4400939941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.50193176269534, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.10043473243713379, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.01170374765060842, + "grad_norm": 0.10692574828863144, + "kept_token_count": 735.8296875, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9934874951839447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006512503232806921, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6828948974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.96348571777344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27730, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.10003017857670785, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.010807002987712621, + "grad_norm": 0.08353360742330551, + "kept_token_count": 736.2734375, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9938331723213196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006166823115199805, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.34047241210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.2902526855469, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.10009787753224372, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.011891529243439437, + "grad_norm": 0.0744226947426796, + "kept_token_count": 736.1984375, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9936198234558106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063801719807088375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8198577880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.49525451660156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27750, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.0994440846145153, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.011445026937872171, + "grad_norm": 0.11016721278429031, + "kept_token_count": 737.021875, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9935731649398803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006426835339516402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6664611816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.6153198242188, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27760, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.0995787188410759, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.00957879377529025, + "grad_norm": 0.10204508900642395, + "kept_token_count": 736.8578125, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9936747550964355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006325238011777401, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9322967529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.35203857421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27770, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.10032265037298202, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.01164782871492207, + "grad_norm": 0.09710565209388733, + "kept_token_count": 736.159375, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9938754856586456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061245298013091086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1949798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.23667297363284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27780, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09965729787945747, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.011757027404382826, + "grad_norm": 0.07809394598007202, + "kept_token_count": 736.859375, + "learning_rate": 4.433002189697938e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9935244619846344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00647554025053978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4675720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1631683349609, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27790, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.09977515265345574, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.012360356468707322, + "grad_norm": 0.12303895503282547, + "kept_token_count": 736.5359375, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9938364386558532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061635770834982395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8915283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1268280029297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27800, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.09935586079955101, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.011383904423564673, + "grad_norm": 0.09778118878602982, + "kept_token_count": 736.8453125, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9935410737991333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006458927690982818, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.02614135742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.47313232421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27810, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.0994655504822731, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.01272015143185854, + "grad_norm": 0.07766775041818619, + "kept_token_count": 736.8484375, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9939531028270722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006046881061047315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.97177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.14202575683595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27820, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.10026350319385528, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.012117554433643818, + "grad_norm": 0.09078269451856613, + "kept_token_count": 736.203125, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9931450068950654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006854991056025028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.34014587402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.08099060058595, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27830, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.10067601576447487, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.009694784414023161, + "grad_norm": 0.1569584608078003, + "kept_token_count": 735.784375, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9928632736206054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007136726006865502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.07110900878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2595153808594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27840, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.09983925074338913, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.010769392177462579, + "grad_norm": 0.06381253153085709, + "kept_token_count": 736.65625, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9936731398105622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006326869316399098, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.956689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.74293823242186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.10076023191213608, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.009993082424625755, + "grad_norm": 0.08913615345954895, + "kept_token_count": 736.0609375, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9936643064022064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006335692573338747, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.61785583496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.71669921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27860, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.09977725222706794, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.011562168411910535, + "grad_norm": 0.1080927699804306, + "kept_token_count": 736.534375, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.993729418516159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006270591169595718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2422302246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.23203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27870, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.10020631477236748, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.012173734419047833, + "grad_norm": 0.0724097192287445, + "kept_token_count": 736.196875, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9936547398567199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006345263123512268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.71025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.4469940185547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27880, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.09994837343692779, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.012787415087223053, + "grad_norm": 0.12645570933818817, + "kept_token_count": 736.1359375, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0161, + "mgd_cos_sim": 0.9934595048427581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065404891036450865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6537750244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.75068359375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27890, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.09972404539585114, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.012031112564727664, + "grad_norm": 0.11802731454372406, + "kept_token_count": 736.7109375, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9936424851417541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00635750824585557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.11448059082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6296661376953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27900, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.0996913380920887, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.012842089124023914, + "grad_norm": 0.09964343905448914, + "kept_token_count": 736.471875, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9939583241939545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006041694059967994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9671569824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.08745727539065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27910, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.10075043886899948, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.01315389471128583, + "grad_norm": 0.10359892249107361, + "kept_token_count": 735.6203125, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0163, + "mgd_cos_sim": 0.9937866449356079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00621334845200181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.58867797851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5915924072266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27920, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.10001547113060952, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.010428335657343269, + "grad_norm": 0.12152186781167984, + "kept_token_count": 736.4109375, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9933776378631591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006622355058789253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0254211425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.21143493652346, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27930, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.09988555759191513, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.011250771675258875, + "grad_norm": 0.09495505690574646, + "kept_token_count": 736.55, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9937076270580292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006292378157377243, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0624206542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.28728942871095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.1009369969367981, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.01040961598046124, + "grad_norm": 0.13933145999908447, + "kept_token_count": 735.5234375, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9939993500709534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006000650022178889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.99058227539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.52308654785156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27950, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.10001500621438027, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.011426523327827454, + "grad_norm": 0.11878763139247894, + "kept_token_count": 736.2859375, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9937848269939422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006215190980583429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4896270751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2681518554688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27960, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.09967701435089112, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.010705593088641763, + "grad_norm": 0.10244869440793991, + "kept_token_count": 736.7078125, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.993011873960495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069881191477179526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.80060119628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.44251403808596, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.10007426142692566, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.01132256155833602, + "grad_norm": 0.07880118489265442, + "kept_token_count": 736.15625, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9939594268798828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006040572375059128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1001220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.4452850341797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27980, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.0994477741420269, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.012494934163987637, + "grad_norm": 0.07558690011501312, + "kept_token_count": 736.88125, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9936882495880127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311742775142193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7740875244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8284515380859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 27990, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.10040893331170082, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.01189256045036018, + "grad_norm": 0.08258526772260666, + "kept_token_count": 736.1859375, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9933505475521087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006649440806359053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.71196899414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5021179199219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28000, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.10012016966938972, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.013149944366887212, + "grad_norm": 0.08094315230846405, + "kept_token_count": 736.3453125, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0162, + "mgd_cos_sim": 0.9938813269138336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006118682585656643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.81654357910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.85796508789065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28010, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.09990426450967789, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.011708920309320092, + "grad_norm": 0.09169761091470718, + "kept_token_count": 736.503125, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9939657330513001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060342707671225074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.88584594726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.89512329101564, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10014484748244286, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.00935652139596641, + "grad_norm": 0.10207684338092804, + "kept_token_count": 736.3828125, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9937247157096862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006275282613933087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.05491943359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.81121520996095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28030, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.09944088235497475, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.012045622011646629, + "grad_norm": 0.11318454146385193, + "kept_token_count": 736.84375, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9941569626331329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005843029823154211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4405059814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0488006591797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28040, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.09943230450153351, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.011950607970356941, + "grad_norm": 0.06771308183670044, + "kept_token_count": 736.9734375, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9937564849853515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006243508122861385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1878234863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1587768554688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28050, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.10009703785181046, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.011771762929856776, + "grad_norm": 0.12197085469961166, + "kept_token_count": 736.3484375, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9938585817813873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006141413003206253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3949035644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9462829589844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28060, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.09965161606669426, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.009812826337292791, + "grad_norm": 0.07761909067630768, + "kept_token_count": 736.88125, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.993466067314148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533932685852051, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6258972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.78147277832034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28070, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.10022503137588501, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.012406070670112967, + "grad_norm": 0.15289995074272156, + "kept_token_count": 736.0453125, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9934678852558136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006532107200473547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0978088378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.4034790039062, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28080, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.1004213660955429, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.012360705947503447, + "grad_norm": 0.1418500691652298, + "kept_token_count": 736.0796875, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9940933465957642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005906670261174441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8365753173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0496032714844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28090, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.10019653812050819, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.010538377426564693, + "grad_norm": 0.11125418543815613, + "kept_token_count": 736.2765625, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9935734033584595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064265959896147255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0448272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.47154846191404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28100, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.10024360567331314, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.013961584260687231, + "grad_norm": 0.11599977314472198, + "kept_token_count": 736.225, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0173, + "mgd_cos_sim": 0.9932950139045715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006704986467957497, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5747528076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.13060607910154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28110, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.10120203718543053, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.01100306841544807, + "grad_norm": 0.09189656376838684, + "kept_token_count": 735.3734375, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9937396168708801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260375306010246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1487243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.97629699707034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28120, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.10022464096546173, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.011693833069875837, + "grad_norm": 0.06667455285787582, + "kept_token_count": 736.4609375, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9939796268939972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006020370498299598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.49712524414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8127838134766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28130, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.1003647655248642, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.011889946553856135, + "grad_norm": 0.05605016648769379, + "kept_token_count": 736.2359375, + "learning_rate": 3.14514360460183e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9937053859233856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006294623669236899, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0068725585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4980194091797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1006819486618042, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.011493761697784067, + "grad_norm": 0.0994947999715805, + "kept_token_count": 735.89375, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9934900462627411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006509948428720236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0166900634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.43645629882815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28150, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.09995903298258782, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.011783417221158743, + "grad_norm": 0.09473344683647156, + "kept_token_count": 736.6046875, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9934625446796417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537452712655068, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9368621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9994689941406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28160, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.1000609576702118, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.011852033855393528, + "grad_norm": 0.08142825216054916, + "kept_token_count": 736.340625, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9935070812702179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006492916867136956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.63246459960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.84794921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28170, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10012673363089561, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.011250345921143889, + "grad_norm": 0.15760594606399536, + "kept_token_count": 736.475, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9937601208686828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006239884067326784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6532379150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6755798339844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.10023985430598259, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.011918805586174131, + "grad_norm": 0.10136435180902481, + "kept_token_count": 735.984375, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9941573679447174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005842626094818115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2886535644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 401.1141387939453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28190, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.09955063089728355, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.00859206598252058, + "grad_norm": 0.07797444611787796, + "kept_token_count": 736.71875, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9938496053218842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061503957025706765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.18421630859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.68169860839845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.10005754232406616, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.010176214016973972, + "grad_norm": 0.08341426402330399, + "kept_token_count": 736.4640625, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9936021745204926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006397831533104181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0759246826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9774627685547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28210, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09993281438946724, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.010147558804601432, + "grad_norm": 0.10052095353603363, + "kept_token_count": 736.971875, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9938080966472626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006191908661276102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.31145629882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4724395751953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28220, + "valid_token_count": 818.796875 + }, + { + "actual_token_mask_ratio": 0.0992509201169014, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.008874556748196483, + "grad_norm": 0.08302818983793259, + "kept_token_count": 737.1046875, + "learning_rate": 2.849237816642192e-07, + "loss": 0.012, + "mgd_cos_sim": 0.993802136182785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061978639103472235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3232116699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.18182983398435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28230, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.09969265386462212, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.011394352186471224, + "grad_norm": 0.09238695353269577, + "kept_token_count": 736.8078125, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9937807500362397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006219247356057167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3045227050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3490264892578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28240, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.10017496645450592, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.011533160088583827, + "grad_norm": 0.09265512973070145, + "kept_token_count": 736.1921875, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9936968088150024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006303199473768473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.05579833984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.6749542236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28250, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.09965725243091583, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.012408106587827206, + "grad_norm": 0.0851581022143364, + "kept_token_count": 736.6125, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9938402891159057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159693188965321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1048522949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5283905029297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28260, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.10057470574975014, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.01211774805560708, + "grad_norm": 0.13801002502441406, + "kept_token_count": 736.0453125, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.993718832731247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006281167734414339, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.99932250976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.82249450683594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28270, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.10032088235020638, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.01156515572220087, + "grad_norm": 0.10375776141881943, + "kept_token_count": 735.89375, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9935089826583863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006491025537252426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.16074523925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.185498046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28280, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.1007484771311283, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.01094614090397954, + "grad_norm": 0.08038922399282455, + "kept_token_count": 735.83125, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9937081694602966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006291830632835627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 393.501708984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28290, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.10011413171887398, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.011027896171435713, + "grad_norm": 0.07669621706008911, + "kept_token_count": 736.259375, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9935891568660736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006410852074623108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.31199035644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2256591796875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28300, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09995703771710396, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.013982642255723477, + "grad_norm": 0.10541704297065735, + "kept_token_count": 736.6234375, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0172, + "mgd_cos_sim": 0.9935669779777527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433029565960169, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5867004394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7797119140625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28310, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.09992983117699623, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.010414921492338181, + "grad_norm": 0.10424746572971344, + "kept_token_count": 736.1359375, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9938565790653229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006143428757786751, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.24152221679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.71050720214845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28320, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.10033118799328804, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.009979404136538505, + "grad_norm": 0.11457005143165588, + "kept_token_count": 736.078125, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9936760663986206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006323950085788965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5382568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8014831542969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28330, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.10043695345520973, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.011890712892636657, + "grad_norm": 0.1555614173412323, + "kept_token_count": 736.1625, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9933715105056763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006628483440726996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.93108825683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.70435485839846, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28340, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.09910672083497048, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.010348658869042992, + "grad_norm": 0.07028880715370178, + "kept_token_count": 737.4125, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9937560319900512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006243966240435839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.69692077636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.31035766601565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28350, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.1000501811504364, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.011700066365301609, + "grad_norm": 0.08105152100324631, + "kept_token_count": 736.1953125, + "learning_rate": 2.447394127561181e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9934683501720428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531664915382862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6425048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5235534667969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28360, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.10011380836367607, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.009987909626215697, + "grad_norm": 0.10792769491672516, + "kept_token_count": 736.1078125, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9932521045207977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747898366302252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.14118041992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.79100646972654, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28370, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.10037028789520264, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.011328256176784634, + "grad_norm": 0.08200261741876602, + "kept_token_count": 735.9, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9934279203414917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006572072673588991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.29765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9807434082031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28380, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.09889486581087112, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.011468585999682545, + "grad_norm": 0.12229467928409576, + "kept_token_count": 737.340625, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9942714452743531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005728568974882364, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0046752929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.61025085449216, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28390, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.10065597966313362, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.01011050008237362, + "grad_norm": 0.10787579417228699, + "kept_token_count": 736.0359375, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9932993292808533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006700679752975702, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.43756408691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.323291015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28400, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.10004453286528588, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.011075247218832374, + "grad_norm": 0.07095631957054138, + "kept_token_count": 736.5, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9937124967575073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006287503521889448, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8422821044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.80727233886716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28410, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.1002326637506485, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.009620604850351811, + "grad_norm": 0.07387954741716385, + "kept_token_count": 736.509375, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9938990473747253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006100963894277811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.48667297363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5476501464844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28420, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.1009857676923275, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.010502007137984038, + "grad_norm": 0.07787330448627472, + "kept_token_count": 735.709375, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9938135921955109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006186420284211636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.68256225585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0861083984375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.09989034086465835, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.009907825710251927, + "grad_norm": 0.1283477246761322, + "kept_token_count": 736.6953125, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.993710321187973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006289663352072239, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.11607666015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3135040283203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28440, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.10043888315558433, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.009891238203272224, + "grad_norm": 0.09444668143987656, + "kept_token_count": 736.2515625, + "learning_rate": 2.186939384802361e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9938150405883789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061849694699049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3361083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2842254638672, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.09946812391281128, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.009884432563558221, + "grad_norm": 0.0979289785027504, + "kept_token_count": 737.1109375, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9936298906803132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006370100192725658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2899200439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.97828674316406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28460, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.09957564175128937, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.011720434576272965, + "grad_norm": 0.09689570963382721, + "kept_token_count": 736.590625, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9936742305755615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006325766071677208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.72471008300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.14374084472655, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28470, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.10030920058488846, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.011802881117910146, + "grad_norm": 0.09796441346406937, + "kept_token_count": 736.25625, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9936267018318177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006373296957463026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8246704101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.81369934082034, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.09948616251349449, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.00942633654922247, + "grad_norm": 0.10492809116840363, + "kept_token_count": 736.7703125, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0128, + "mgd_cos_sim": 0.9933430790901184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006656916998326779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0102142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.38063659667966, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28490, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09978302717208862, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.013468960905447603, + "grad_norm": 0.11197301000356674, + "kept_token_count": 736.74375, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0169, + "mgd_cos_sim": 0.9931663036346435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006833694782108069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.73089294433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.7764373779297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28500, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.1007970318198204, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.010792905976995826, + "grad_norm": 0.07174822688102722, + "kept_token_count": 735.6390625, + "learning_rate": 2.021383606975502e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9936531126499176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063468881882727144, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2917144775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4274017333984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28510, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.09980056807398796, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.009614945435896517, + "grad_norm": 0.06481709331274033, + "kept_token_count": 736.8296875, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0128, + "mgd_cos_sim": 0.9936586916446686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341324932873249, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3752105712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.13054809570315, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28520, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.09966370686888695, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.009641213342547417, + "grad_norm": 0.07110536098480225, + "kept_token_count": 736.6375, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9941372036933899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005862781219184399, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.77288513183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0456146240234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28530, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.1002614140510559, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.010848186630755662, + "grad_norm": 0.09514667838811874, + "kept_token_count": 736.33125, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9939011394977569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060988604091107845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.29998779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4544342041016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28540, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.10033142790198327, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.010441784746944904, + "grad_norm": 0.079281285405159, + "kept_token_count": 736.0234375, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9935186386108399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064813709817826744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1047637939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0499969482422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28550, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.1005842626094818, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.00993386316113174, + "grad_norm": 0.09559130668640137, + "kept_token_count": 736.15625, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9940982520580292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005901749338954687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.20298767089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.4858703613281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28560, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.10000855773687363, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.012298029009252786, + "grad_norm": 0.06908927112817764, + "kept_token_count": 736.55, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9936560750007629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006343921460211277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9671234130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.02720031738284, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28570, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.10044312998652458, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.013883135607466101, + "grad_norm": 0.11948802322149277, + "kept_token_count": 736.0484375, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.017, + "mgd_cos_sim": 0.9938600122928619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006139997951686382, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0482208251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.59601440429685, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28580, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.09955082237720489, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01056953207589686, + "grad_norm": 0.12935486435890198, + "kept_token_count": 736.8, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9937157452106475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006284248270094395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3654449462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.68601684570314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28590, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09973454847931862, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.012332075927406549, + "grad_norm": 0.0866207554936409, + "kept_token_count": 736.93125, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9935413062572479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006458699982613325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6336944580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.37529907226565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28600, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.10031399577856064, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.00989056983962655, + "grad_norm": 0.12239670753479004, + "kept_token_count": 736.3625, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9937477886676789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006252213660627603, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.76817016601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7724212646484, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.09939521476626396, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.011216256627812981, + "grad_norm": 0.06135369464755058, + "kept_token_count": 736.990625, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9938993215560913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00610068803653121, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5915283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0915924072266, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28620, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.1006226360797882, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.011578019894659519, + "grad_norm": 0.09524048864841461, + "kept_token_count": 736.0234375, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9940125226974488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005987473204731942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.20472412109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3082763671875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28630, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.10007538497447968, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.011687015835195779, + "grad_norm": 0.1141725406050682, + "kept_token_count": 736.4328125, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9934053778648376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006594625581055879, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6110046386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4206909179687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28640, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.1000428818166256, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.009714542981237173, + "grad_norm": 0.08965696394443512, + "kept_token_count": 736.2234375, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9937090814113617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062909252010285854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9419738769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.3671417236328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28650, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.10023910850286484, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.01108078397810459, + "grad_norm": 0.11124150454998016, + "kept_token_count": 735.9078125, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9934206128120422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006579395942389965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.57261047363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5684539794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28660, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.09995757937431335, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.012071332428604364, + "grad_norm": 0.09059105068445206, + "kept_token_count": 736.228125, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9941130042076111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005886990390717983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.35157775878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0676971435547, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28670, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.09917444586753846, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.008978047594428062, + "grad_norm": 0.09903786331415176, + "kept_token_count": 737.2828125, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9939872026443481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060128005221486095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0489532470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.5475555419922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28680, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.09971000254154205, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.011107231071218849, + "grad_norm": 0.0956467017531395, + "kept_token_count": 736.721875, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9937607944011688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062392082996666435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.20542297363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1412078857422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28690, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.09966968968510628, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.010874781897291541, + "grad_norm": 0.1253260374069214, + "kept_token_count": 736.8875, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9938100874423981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006189905852079391, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.05914611816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0589630126953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28700, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.09920430034399033, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.010422420594841242, + "grad_norm": 0.09431719779968262, + "kept_token_count": 736.9890625, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9937554478645325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062445517629385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2115966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.1985137939453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28710, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09995889887213708, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.011980696627870203, + "grad_norm": 0.08772657066583633, + "kept_token_count": 736.4875, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9933572590351105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642729695886374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2648529052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8952697753906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28720, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.10064183846116066, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.009849668247625232, + "grad_norm": 0.0576401986181736, + "kept_token_count": 736.290625, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9937876105308533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006212402787059546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.49294128417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.35121459960936, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28730, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.10004616677761077, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.010699600214138627, + "grad_norm": 0.0802454799413681, + "kept_token_count": 736.3984375, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9939808070659637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006019203644245863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4090209960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.32831420898435, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.10039503574371338, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.012187560088932514, + "grad_norm": 0.1016591489315033, + "kept_token_count": 735.9796875, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9939152121543884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060847818851470946, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.68944091796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9265563964844, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28750, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.0999014213681221, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.01227872553281486, + "grad_norm": 0.07826707512140274, + "kept_token_count": 736.4234375, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9933977365493775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006602250412106514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.04978637695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1634552001953, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28760, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.09966024309396744, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.010507737845182418, + "grad_norm": 0.10308553278446198, + "kept_token_count": 737.003125, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9936377167701721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006362292170524597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0863098144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0536376953125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28770, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.09921810626983643, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.01211459874175489, + "grad_norm": 0.07501721382141113, + "kept_token_count": 737.1625, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9935512185096741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448788940906525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9567443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8623565673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28780, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.10079526975750923, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.010052466858178377, + "grad_norm": 0.12784212827682495, + "kept_token_count": 735.7796875, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9937008142471313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006299187988042831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8206512451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5752227783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28790, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.10018475949764252, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.010661547491326927, + "grad_norm": 0.10342606902122498, + "kept_token_count": 736.16875, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9934455335140229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006554474681615829, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4370422363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8860229492187, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.10005637779831886, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.011930128186941147, + "grad_norm": 0.11157364398241043, + "kept_token_count": 736.3125, + "learning_rate": 1.290835601087864e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9937604069709778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062395835295319555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.44053649902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7839660644531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28810, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.0996038168668747, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.012965609319508076, + "grad_norm": 0.09725353866815567, + "kept_token_count": 736.8671875, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0161, + "mgd_cos_sim": 0.9936812281608581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006318783853203058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.581884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.00172729492186, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.0999984011054039, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.012384749436751007, + "grad_norm": 0.1046089455485344, + "kept_token_count": 736.4671875, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9934942543506622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065057573840022085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.68046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.8860717773438, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.10039573088288307, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.011131467623636126, + "grad_norm": 0.10896599292755127, + "kept_token_count": 736.434375, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9936440587043762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006355923600494862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6536010742187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.310546875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28840, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.09984337985515594, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.010991841554641724, + "grad_norm": 0.08193439990282059, + "kept_token_count": 736.4015625, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9936979234218597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006302083842456341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.61499938964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.9841003417969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28850, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.10048122927546502, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.011562404735013843, + "grad_norm": 0.1646028459072113, + "kept_token_count": 736.0859375, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9939493775367737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006050639227032662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1802673339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.44578247070314, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28860, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.09980234131217003, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.01473592733964324, + "grad_norm": 0.1305626779794693, + "kept_token_count": 736.9859375, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.018, + "mgd_cos_sim": 0.9933803915977478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006619609985500574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.18614807128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.28406677246096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28870, + "valid_token_count": 818.6953125 + }, + { + "actual_token_mask_ratio": 0.09984687566757203, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.011762956483289599, + "grad_norm": 0.11665770411491394, + "kept_token_count": 736.8015625, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9935561656951905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443838216364383, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1118133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4071746826172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28880, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.09929931610822677, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.010344111174345017, + "grad_norm": 0.09972125291824341, + "kept_token_count": 736.8890625, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9936570763587952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006342922244220972, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0017150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9057861328125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28890, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10023572817444801, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.012071497086435557, + "grad_norm": 0.08260597288608551, + "kept_token_count": 736.178125, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.993658185005188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341803446412086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.40796813964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.0987243652344, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28900, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.10045020952820778, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.012848594412207603, + "grad_norm": 0.09365081787109375, + "kept_token_count": 736.1890625, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9938102960586548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061896972358226774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9386322021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0514739990234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.09933377429842949, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.010631298227235676, + "grad_norm": 0.11214471608400345, + "kept_token_count": 737.040625, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9939920008182526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008018646389246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1407501220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.96965942382815, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28920, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.09943146333098411, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.011789962090551854, + "grad_norm": 0.1960240751504898, + "kept_token_count": 736.853125, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9936385571956634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006361432466655969, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0035125732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.71458435058594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28930, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10082921460270881, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.010540357837453485, + "grad_norm": 0.08840811252593994, + "kept_token_count": 735.7953125, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9938294291496277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006170576997101307, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.45203552246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.1682891845703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28940, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.10042092204093933, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.013875927869230508, + "grad_norm": 0.08460728824138641, + "kept_token_count": 736.1265625, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.017, + "mgd_cos_sim": 0.9936687886714936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006331227533519268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2106994628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.72747802734375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28950, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.09934222176671029, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.012528640683740377, + "grad_norm": 0.08296994864940643, + "kept_token_count": 736.7109375, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.9938749194145202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006125072669237852, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.931396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.25791015625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28960, + "valid_token_count": 817.971875 + }, + { + "actual_token_mask_ratio": 0.09998801052570343, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.010740897571668028, + "grad_norm": 0.09606651216745377, + "kept_token_count": 736.5265625, + "learning_rate": 9.676562902984965e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9935103356838226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064896678552031515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4344787597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.588623046875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28970, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.09958281889557838, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.011621102131903172, + "grad_norm": 0.11972647160291672, + "kept_token_count": 736.5921875, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9937865793704986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006213418953120708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8826629638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0698547363281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28980, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.10065422430634499, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.010259888647124172, + "grad_norm": 0.07821501791477203, + "kept_token_count": 735.575, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0137, + "mgd_cos_sim": 0.9932038366794587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006796165462583303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1829833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.3460968017578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 28990, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.0991701290011406, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.010217801574617625, + "grad_norm": 0.06721667945384979, + "kept_token_count": 736.8765625, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0132, + "mgd_cos_sim": 0.9940609514713288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059390589594841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0197998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3776397705078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29000, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.099959696829319, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.011374887684360147, + "grad_norm": 0.09445217996835709, + "kept_token_count": 736.2, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9934238731861115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006576137244701385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5444610595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.501025390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29010, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.10029519498348236, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.012305710976943374, + "grad_norm": 0.11119294166564941, + "kept_token_count": 736.4484375, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9936273157596588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006372667476534843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3899261474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5552612304688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29020, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.10073161646723747, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.011140748672187329, + "grad_norm": 0.09341875463724136, + "kept_token_count": 735.415625, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9935217916965484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064782162196934225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.827978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.5257629394531, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29030, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.10001453086733818, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.011340353917330504, + "grad_norm": 0.11374463140964508, + "kept_token_count": 736.2515625, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9938514411449433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148545723408461, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.60353088378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.60158081054686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29040, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.09957767203450203, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.012369879893958569, + "grad_norm": 0.06506257504224777, + "kept_token_count": 737.0171875, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.9936232924461365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006376715190708637, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.52565612792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0360382080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29050, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.10011615753173828, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.01223769155330956, + "grad_norm": 0.0978945791721344, + "kept_token_count": 736.0765625, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0154, + "mgd_cos_sim": 0.9935808122158051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064191905781626705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1427276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2537353515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29060, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.09998122528195381, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.011946007888764144, + "grad_norm": 0.11677803099155426, + "kept_token_count": 736.390625, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0152, + "mgd_cos_sim": 0.9935346186161041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006465370021760464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.40724182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1770935058594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29070, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.09960316345095635, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.01096532391384244, + "grad_norm": 0.08947767317295074, + "kept_token_count": 736.9890625, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.99338800907135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612000148743391, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.09754638671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3538513183594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29080, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.10044827237725258, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.009803557721897959, + "grad_norm": 0.08983856439590454, + "kept_token_count": 736.009375, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0129, + "mgd_cos_sim": 0.9938324809074401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006167536973953247, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.87396545410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2669616699219, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29090, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.10062307119369507, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.01120651001110673, + "grad_norm": 0.07298648357391357, + "kept_token_count": 735.63125, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9937760591506958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006223933957517147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7009552001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8203918457031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29100, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.10038491785526275, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.011690439423546195, + "grad_norm": 0.11490995436906815, + "kept_token_count": 736.4015625, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0148, + "mgd_cos_sim": 0.9936939477920532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006306060217320919, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3141693115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8002227783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29110, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.09948528185486794, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.01399578582495451, + "grad_norm": 0.12219049036502838, + "kept_token_count": 736.7625, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0173, + "mgd_cos_sim": 0.9934075593948364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006592445634305477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5582550048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3927795410156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29120, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.09934332072734833, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.01216054568067193, + "grad_norm": 0.07741878926753998, + "kept_token_count": 736.94375, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9933533608913422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006646637059748173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3938262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.86615600585935, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29130, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.1002987079322338, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.012505228165537119, + "grad_norm": 0.07912036031484604, + "kept_token_count": 736.190625, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0157, + "mgd_cos_sim": 0.9936333894729614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366605591028929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.49354248046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.4774169921875, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29140, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.09933843165636062, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.011574071692302823, + "grad_norm": 0.06902025640010834, + "kept_token_count": 737.0546875, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0149, + "mgd_cos_sim": 0.9934219658374787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065780352801084515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7929260253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5399536132812, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29150, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.0998992808163166, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.011096927709877491, + "grad_norm": 0.09013308584690094, + "kept_token_count": 736.3234375, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9940268695354462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059731341898441315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.01958618164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8764953613281, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29160, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.09981977865099907, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.011096274852752686, + "grad_norm": 0.07335411012172699, + "kept_token_count": 736.546875, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9937522768974304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00624773083254695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8287048339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.2753540039063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29170, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.10037180334329605, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.012908054189756513, + "grad_norm": 0.10278933495283127, + "kept_token_count": 736.1796875, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0161, + "mgd_cos_sim": 0.9937062203884125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006293780077248812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7435302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1926544189453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29180, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.10001982375979424, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.0107382882386446, + "grad_norm": 0.09655528515577316, + "kept_token_count": 736.5375, + "learning_rate": 5.989957490296183e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9934479236602783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552072148770094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2914123535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3724700927734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29190, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.10053634867072106, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.009205196658149362, + "grad_norm": 0.08474452793598175, + "kept_token_count": 735.884375, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0122, + "mgd_cos_sim": 0.9939613044261932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00603869715705514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.90451965332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.747900390625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29200, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.10004937201738358, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.011165874172002077, + "grad_norm": 0.14641644060611725, + "kept_token_count": 736.571875, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9940626442432403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005937358178198337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.13564147949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2649322509766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29210, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.09975284039974212, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.011579166911542416, + "grad_norm": 0.09332875907421112, + "kept_token_count": 736.665625, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9939205467700958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006079446617513895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.63779602050784, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29220, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.10126351118087769, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.011175165744498372, + "grad_norm": 0.09267827868461609, + "kept_token_count": 735.3546875, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9938407897949219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00615921076387167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.62181701660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0147735595703, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.0999229110777378, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.011343844793736934, + "grad_norm": 0.062392815947532654, + "kept_token_count": 737.090625, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9936186790466308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381318811327219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.22777099609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.40398864746095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29240, + "valid_token_count": 818.9203125 + }, + { + "actual_token_mask_ratio": 0.09992073476314545, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.010763069149106741, + "grad_norm": 0.09148484468460083, + "kept_token_count": 736.5078125, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.993725198507309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006274794321507215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.47583923339846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2189117431641, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29250, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09991006031632424, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.012158718798309565, + "grad_norm": 0.09216982871294022, + "kept_token_count": 736.5609375, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0152, + "mgd_cos_sim": 0.9939439713954925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060560211539268495, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4354522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0153472900391, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29260, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.09937323704361915, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.01205994114279747, + "grad_norm": 0.11547908186912537, + "kept_token_count": 736.903125, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0151, + "mgd_cos_sim": 0.9938489973545075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006150997430086136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7796691894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.0345916748047, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29270, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.10053427219390869, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.01180070056580007, + "grad_norm": 0.13010302186012268, + "kept_token_count": 736.071875, + "learning_rate": 4.734926068397038e-08, + "loss": 0.015, + "mgd_cos_sim": 0.993668657541275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006331352796405554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4282989501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8528289794922, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29280, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.09996799901127815, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.012030677171424031, + "grad_norm": 0.11221254616975784, + "kept_token_count": 736.171875, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0153, + "mgd_cos_sim": 0.9933957695960999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006604220904409886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5213989257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.1333526611328, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29290, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.10004686042666436, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.010597689310088754, + "grad_norm": 0.1268714964389801, + "kept_token_count": 736.26875, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.9934559166431427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006544088665395975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.25000915527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.8764678955078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29300, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.10004616379737855, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.010813620081171394, + "grad_norm": 0.09610674530267715, + "kept_token_count": 736.5578125, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.99314643740654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006853550579398871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5950500488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.55, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29310, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.09997740536928176, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.013197370851412416, + "grad_norm": 0.11587602645158768, + "kept_token_count": 736.446875, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0163, + "mgd_cos_sim": 0.9937698900699615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006230107508599758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.52219848632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.9242309570312, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29320, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.1001664325594902, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.010079256305471063, + "grad_norm": 0.07784972339868546, + "kept_token_count": 736.278125, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0133, + "mgd_cos_sim": 0.9935701668262482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006429829727858305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.50274658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.98329467773436, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29330, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.10062448531389237, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.012416443368420006, + "grad_norm": 0.09890125691890717, + "kept_token_count": 736.1421875, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9937744677066803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225530896335841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3246673583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.09813232421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.09970488473773002, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.012464509531855582, + "grad_norm": 0.08089800924062729, + "kept_token_count": 736.7359375, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9939111948013306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006088797282427549, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7527191162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2735595703125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.09980519488453865, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.011335856188088656, + "grad_norm": 0.08954238891601562, + "kept_token_count": 736.503125, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9934478342533112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00655217245221138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6616973876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.50750427246095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29360, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.09914900660514832, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.011162058729678393, + "grad_norm": 0.1060391217470169, + "kept_token_count": 737.4765625, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9934722244739532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00652777710929513, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.88018188476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.25202026367185, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29370, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.09967222437262535, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.010490991827100515, + "grad_norm": 0.10614537447690964, + "kept_token_count": 736.853125, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0136, + "mgd_cos_sim": 0.9936913251876831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006308677885681391, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.57664489746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.2732696533203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29380, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.09977945983409882, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.010913105588406325, + "grad_norm": 0.09899257123470306, + "kept_token_count": 736.4515625, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9935933351516724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006406663917005062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1446960449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8750701904297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29390, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.10052296072244644, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.01104487581178546, + "grad_norm": 0.10985400527715683, + "kept_token_count": 735.9609375, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9937226951122284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006277297157794237, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1202453613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.39586486816404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29400, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.10011302083730697, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.010784595040604472, + "grad_norm": 0.07309470325708389, + "kept_token_count": 736.396875, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0137, + "mgd_cos_sim": 0.9940758943557739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005924110021442175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.35682678222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8735382080078, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29410, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.10021033287048339, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.011260661389678716, + "grad_norm": 0.13303108513355255, + "kept_token_count": 736.1359375, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9936933577060699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006306653469800949, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.27467041015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.6872497558594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29420, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.10004144832491875, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.014642396476119757, + "grad_norm": 0.1292024552822113, + "kept_token_count": 736.734375, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0179, + "mgd_cos_sim": 0.9934681475162506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531837396323681, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4636260986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.3058349609375, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29430, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.10047917291522027, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.008232264267280697, + "grad_norm": 0.060922738164663315, + "kept_token_count": 735.8421875, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.9937712728977204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062287417240440845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3034332275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.3185516357422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29440, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.09984485208988189, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.010149088548496365, + "grad_norm": 0.08923035860061646, + "kept_token_count": 736.584375, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0131, + "mgd_cos_sim": 0.9940248966217041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005975113250315189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.63515014648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.78602294921876, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29450, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.10007556453347206, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.013857803959399461, + "grad_norm": 0.1799742877483368, + "kept_token_count": 736.471875, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0171, + "mgd_cos_sim": 0.9934594929218292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006540504842996597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.10679321289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9088409423828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29460, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.09985955283045769, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.010821603704243898, + "grad_norm": 0.08172879368066788, + "kept_token_count": 736.5921875, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9936735808849335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063264347612857815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7729187011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2723022460938, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29470, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.10015313848853111, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.011188203329220414, + "grad_norm": 0.10004279017448425, + "kept_token_count": 736.2046875, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9933576047420501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642391160130501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8812316894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.9761169433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29480, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.0999857246875763, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.01227525770664215, + "grad_norm": 0.09188494831323624, + "kept_token_count": 736.5046875, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0154, + "mgd_cos_sim": 0.9936691701412201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006330851092934608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.62685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1459014892578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29490, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.1003012329339981, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.011973416665568948, + "grad_norm": 0.1011485755443573, + "kept_token_count": 736.0140625, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.993005883693695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069941127672791484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6172607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.53619384765625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29500, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.09950688928365707, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.010969836870208383, + "grad_norm": 0.10801015794277191, + "kept_token_count": 737.04375, + "learning_rate": 2.196485058207298e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9939156174659729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084390822798014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0112701416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1975402832031, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29510, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.10028945580124855, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.011435799207538367, + "grad_norm": 0.08338561654090881, + "kept_token_count": 735.8703125, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9934553921222686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00654460871592164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0142761230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6680908203125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29520, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.09960614368319512, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.010546944150701166, + "grad_norm": 0.05696980655193329, + "kept_token_count": 736.6109375, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0136, + "mgd_cos_sim": 0.993847918510437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006152080558240414, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.69443664550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.8645965576172, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29530, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.09981531351804733, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.011657605599611997, + "grad_norm": 0.08533238619565964, + "kept_token_count": 736.5171875, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0149, + "mgd_cos_sim": 0.9935454368591309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006454560346901417, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.67711181640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.4251312255859, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29540, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.10020874589681625, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.01015861346386373, + "grad_norm": 0.09552876651287079, + "kept_token_count": 736.3734375, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0135, + "mgd_cos_sim": 0.9933517813682556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006648218259215355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2094482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.9117767333984, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29550, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.09909906685352325, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.011521470360457897, + "grad_norm": 0.08986277878284454, + "kept_token_count": 737.2203125, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9936271190643311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637287376448512, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0102020263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.98500061035156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29560, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.09960124269127846, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.010164585150778293, + "grad_norm": 0.09371304512023926, + "kept_token_count": 736.70625, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0133, + "mgd_cos_sim": 0.9936753749847412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006324636470526457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.21861267089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.7047943115234, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29570, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.10029477700591087, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.013623679149895906, + "grad_norm": 0.10572437942028046, + "kept_token_count": 736.3265625, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0167, + "mgd_cos_sim": 0.9937974810600281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006202523969113827, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.41107482910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.18089904785154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29580, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.09917104914784432, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.010692989500239492, + "grad_norm": 0.13927264511585236, + "kept_token_count": 737.2125, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.9935859143733978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006414094287902117, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1810760498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.0890197753906, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29590, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1000267818570137, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.011090382048860193, + "grad_norm": 0.09185998141765594, + "kept_token_count": 736.25625, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9935761749744415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006423817668110132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.22845764160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.6487701416016, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29600, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.10005312636494637, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.010533235128968954, + "grad_norm": 0.11589709669351578, + "kept_token_count": 736.5171875, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0138, + "mgd_cos_sim": 0.9934536159038544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065463785082101825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7338348388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.3420684814453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29610, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.09994308352470398, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.010619988292455673, + "grad_norm": 0.07662086188793182, + "kept_token_count": 736.49375, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.9934242010116577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006575818825513125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4760772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3303619384766, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29620, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.09984332770109176, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.012182230968028307, + "grad_norm": 0.09259507805109024, + "kept_token_count": 736.509375, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0153, + "mgd_cos_sim": 0.9937846839427948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006215326674282551, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9435638427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6382690429688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29630, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.09965587109327316, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.012292200699448586, + "grad_norm": 0.10419807583093643, + "kept_token_count": 736.609375, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0154, + "mgd_cos_sim": 0.9937721073627472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006227884534746408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2053253173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.21880798339845, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29640, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.10046715587377548, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.01144829816184938, + "grad_norm": 0.10086021572351456, + "kept_token_count": 736.196875, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9936048448085785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006395152676850558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.84283752441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5668518066406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29650, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.09955629855394363, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.010581636056303977, + "grad_norm": 0.09757890552282333, + "kept_token_count": 736.9453125, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0135, + "mgd_cos_sim": 0.9940792858600617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005920697934925556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9733093261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3607513427734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29660, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.10022328868508339, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.009175978787243366, + "grad_norm": 0.09776518493890762, + "kept_token_count": 736.375, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0122, + "mgd_cos_sim": 0.9939903378486633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006009668204933405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1454162597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.6251678466797, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29670, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.1002065122127533, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.01141886510886252, + "grad_norm": 0.09977901726961136, + "kept_token_count": 736.2859375, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0146, + "mgd_cos_sim": 0.9936188519001007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381148751825094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1733703613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.78172607421874, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29680, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.09918591305613518, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.012254140991717576, + "grad_norm": 0.09652413427829742, + "kept_token_count": 737.1, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0153, + "mgd_cos_sim": 0.9938941955566406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006105802394449711, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6833526611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.8178283691406, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.09993158057332038, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.01251807170920074, + "grad_norm": 0.14017511904239655, + "kept_token_count": 736.6046875, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0157, + "mgd_cos_sim": 0.9936425626277924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006357429921627045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.48447875976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.51976318359374, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29700, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.10019492655992508, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.010934907477349043, + "grad_norm": 0.08219720423221588, + "kept_token_count": 736.240625, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0142, + "mgd_cos_sim": 0.9934568703174591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006543128658086062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7033996582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.35613708496095, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.10004453733563423, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.010422473540529609, + "grad_norm": 0.07662023603916168, + "kept_token_count": 736.4296875, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0134, + "mgd_cos_sim": 0.994041633605957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005958371143788099, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.62745056152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.11383056640625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29720, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.09969712272286416, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.011071179155260324, + "grad_norm": 0.09817676246166229, + "kept_token_count": 736.7203125, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0141, + "mgd_cos_sim": 0.9939180076122284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006082000955939293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2049041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.7607940673828, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.10027948692440987, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.009596822038292884, + "grad_norm": 0.09587761759757996, + "kept_token_count": 736.4578125, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0126, + "mgd_cos_sim": 0.9939134240150451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006086582131683826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.29909057617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.8122924804687, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29740, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.09986864253878594, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.01202531261369586, + "grad_norm": 0.1355157494544983, + "kept_token_count": 736.7828125, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0153, + "mgd_cos_sim": 0.9933999121189118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00660008629783988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.14171447753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.18834533691404, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29750, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.09953586757183075, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.011839007819071412, + "grad_norm": 0.09377831965684891, + "kept_token_count": 736.925, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.015, + "mgd_cos_sim": 0.9936881184577941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311880610883236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.818896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.14092712402345, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29760, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.10031013786792756, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.011787468567490578, + "grad_norm": 0.0982896015048027, + "kept_token_count": 736.1625, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.015, + "mgd_cos_sim": 0.9935502529144287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449756398797035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1738220214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.5095977783203, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.09952873066067695, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.01245176033116877, + "grad_norm": 0.089340440928936, + "kept_token_count": 736.85625, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0157, + "mgd_cos_sim": 0.993490731716156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006509256921708584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.12615966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.1330078125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29780, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.0992976762354374, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.01094751600176096, + "grad_norm": 0.08135160803794861, + "kept_token_count": 736.921875, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0141, + "mgd_cos_sim": 0.9937329649925232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006267036683857441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.86395568847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 394.4658966064453, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29790, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.10014223977923394, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.0111448525916785, + "grad_norm": 0.0987236425280571, + "kept_token_count": 736.0625, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0142, + "mgd_cos_sim": 0.9938342332839966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00616575675085187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3643524169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3804565429688, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29800, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.10020114704966546, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.011995617393404245, + "grad_norm": 0.09865804016590118, + "kept_token_count": 736.2546875, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0151, + "mgd_cos_sim": 0.9937350988388062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006264912802726031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8010955810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.32680358886716, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29810, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.09914479032158852, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.011109900940209628, + "grad_norm": 0.10273806005716324, + "kept_token_count": 736.9796875, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0143, + "mgd_cos_sim": 0.9936006963253021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006399314384907484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.23583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.3979919433594, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29820, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.1004638522863388, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.011764728557318449, + "grad_norm": 0.10720271617174149, + "kept_token_count": 736.14375, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0148, + "mgd_cos_sim": 0.9939217925071716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006078213639557362, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.35932006835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.5863891601563, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29830, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.09983909651637077, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.010739546222612261, + "grad_norm": 0.07547091692686081, + "kept_token_count": 736.4875, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0138, + "mgd_cos_sim": 0.9938704013824463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006129605695605278, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.60361938476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.03539123535154, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29840, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.09905461221933365, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.012200734298676252, + "grad_norm": 0.08870716392993927, + "kept_token_count": 737.1015625, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0154, + "mgd_cos_sim": 0.9935811936855317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418822612613439, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6687347412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.7703460693359, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29850, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.1001477763056755, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.01267838841304183, + "grad_norm": 0.06814493238925934, + "kept_token_count": 736.559375, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0156, + "mgd_cos_sim": 0.9941889524459839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005811043176800012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4719268798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.95033264160156, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29860, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.1005486786365509, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.010828649252653122, + "grad_norm": 0.08625084161758423, + "kept_token_count": 735.8515625, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0139, + "mgd_cos_sim": 0.993849104642868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006150894425809383, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.61903381347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.28179321289065, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29870, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.10057018846273422, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.011572373192757368, + "grad_norm": 0.10559586435556412, + "kept_token_count": 735.8703125, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0147, + "mgd_cos_sim": 0.9937832713127136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006216722633689642, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1302215576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.71730346679686, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29880, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.09977753385901451, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.012605596520006656, + "grad_norm": 0.09545932710170746, + "kept_token_count": 736.5609375, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0156, + "mgd_cos_sim": 0.9939137101173401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006086292862892151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4097381591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.06035766601565, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29890, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.09915561452507973, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.011449132673442363, + "grad_norm": 0.07017812877893448, + "kept_token_count": 737.3390625, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0147, + "mgd_cos_sim": 0.9935375034809113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006462492421269417, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.15008239746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.0585906982422, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29900, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.10024083405733109, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.010659110592678189, + "grad_norm": 0.07538051158189774, + "kept_token_count": 736.440625, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0137, + "mgd_cos_sim": 0.9938265204429626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006173483375459909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1699188232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 396.52589111328126, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29910, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.09967379122972489, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.011441789101809263, + "grad_norm": 0.08620873093605042, + "kept_token_count": 736.58125, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0145, + "mgd_cos_sim": 0.9938288867473603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171106919646263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6573974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.80164489746096, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29920, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.10000214800238609, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.014477906096726657, + "grad_norm": 0.17647366225719452, + "kept_token_count": 736.3296875, + "learning_rate": 4.593944767161684e-10, + "loss": 0.018, + "mgd_cos_sim": 0.9930183827877045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006981617864221334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3284484863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 395.471728515625, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29930, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.10048317015171052, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.010971714789047838, + "grad_norm": 0.08886290341615677, + "kept_token_count": 736.065625, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0142, + "mgd_cos_sim": 0.9935690224170685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006430978607386351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.88714904785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.2512939453125, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29940, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.10028431192040443, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.012041822634637355, + "grad_norm": 0.13367873430252075, + "kept_token_count": 736.2109375, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0153, + "mgd_cos_sim": 0.9935831904411316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00641680397093296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7416656494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.68846130371094, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.0999649703502655, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.010723659303039312, + "grad_norm": 0.09319771826267242, + "kept_token_count": 736.446875, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0138, + "mgd_cos_sim": 0.9937584459781647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00624155281111598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.54661254882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 398.2774139404297, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29960, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.09964059814810752, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.010958096524700523, + "grad_norm": 0.10332333296537399, + "kept_token_count": 736.5421875, + "learning_rate": 8.757784487611886e-11, + "loss": 0.014, + "mgd_cos_sim": 0.9938419103622437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061580948531627655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.33766784667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.7636352539063, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29970, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.0999683491885662, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.010184232844039798, + "grad_norm": 0.14684680104255676, + "kept_token_count": 736.2234375, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0133, + "mgd_cos_sim": 0.9936775624752044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006322437897324562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.467431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 399.9525421142578, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29980, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.09984266906976699, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.010020628897473216, + "grad_norm": 0.09117042273283005, + "kept_token_count": 736.7296875, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0129, + "mgd_cos_sim": 0.9941514372825623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058485673740506176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4102020263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 397.5641784667969, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 29990, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.09966549947857857, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.009604488965123892, + "grad_norm": 0.08894732594490051, + "kept_token_count": 736.984375, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0127, + "mgd_cos_sim": 0.9937545776367187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006245433725416661, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9737274169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 400.2211029052734, + "mgd_token_mask_ratio_cfg": 0.10000000149011612, + "step": 30000, + "valid_token_count": 818.5625 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.01864964050129056, + "train_runtime": 59155.7177, + "train_samples_per_second": 64.913, + "train_steps_per_second": 0.507 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..88051fbddd5272d940be596deafdb50501d57774 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.1/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e36e4efc7535170dfa141d25fdb685596fc171a8fe10bafd33f03dfb3c37f9b +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/config.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e76e5434e4901378a9c3da0df6b159622cfdfe38 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": true, + "mgd_fm_loss_weight": 0.0, + "mgd_loss_weight": 1.0, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": null, + "mgd_loss_weight_start": 0.05, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.2, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/experiment_cfg/metadata.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c9509993e12e3bc3b291d1a0398318e256c8a2b7 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0c469d4502bf8d706dbc1069352deea7bc4b74f5 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7546a323a674ff18936f5b04a14c992b8b4388c0d58416e10a286ef15fdec8e8 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/trainer_state.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..fcb7b5942198555c94e8ddc02620b88f1cb79fdf --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.20007037222385407, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.01689767502248287, + "grad_norm": 1.4725685119628906, + "kept_token_count": 654.7640625, + "learning_rate": 1.8e-07, + "loss": 0.9537, + "mgd_cos_sim": 0.046308889985084534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.9536911129951477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.51049461364746, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8511077880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.20078185945749283, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.015974902361631394, + "grad_norm": 1.5413624048233032, + "kept_token_count": 654.028125, + "learning_rate": 3.8e-07, + "loss": 0.9473, + "mgd_cos_sim": 0.0526777770370245, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.9473222136497498, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.260022926330567, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11327209472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19907019883394242, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.017956522479653358, + "grad_norm": 1.4850841760635376, + "kept_token_count": 655.4609375, + "learning_rate": 5.800000000000001e-07, + "loss": 0.9422, + "mgd_cos_sim": 0.05775653384625912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.9422434806823731, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.1360933303833, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66038208007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 30, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19951053112745284, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.01390380710363388, + "grad_norm": 1.4874178171157837, + "kept_token_count": 655.215625, + "learning_rate": 7.799999999999999e-07, + "loss": 0.9305, + "mgd_cos_sim": 0.06951100714504718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.9304889917373658, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.290578842163086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8821044921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 40, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.20040583312511445, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.014863232243806123, + "grad_norm": 1.5257518291473389, + "kept_token_count": 654.2734375, + "learning_rate": 9.8e-07, + "loss": 0.9152, + "mgd_cos_sim": 0.08484395891427994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.9151560485363006, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.005090522766114, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.14263305664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 50, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20026148855686188, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.016725091636180876, + "grad_norm": 1.4883660078048706, + "kept_token_count": 654.2875, + "learning_rate": 1.18e-06, + "loss": 0.8956, + "mgd_cos_sim": 0.10438237488269805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.8956176340579987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 24.97743663787842, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6015106201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 60, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20076031535863875, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.019012610334903, + "grad_norm": 1.5131704807281494, + "kept_token_count": 654.1015625, + "learning_rate": 1.38e-06, + "loss": 0.8742, + "mgd_cos_sim": 0.12581645995378493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.8741835296154022, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 24.539094161987304, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2132232666016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 70, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20059806108474731, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.015283714979887009, + "grad_norm": 1.5971295833587646, + "kept_token_count": 654.14375, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.8431, + "mgd_cos_sim": 0.15691855549812317, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.8430814325809479, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 23.773989868164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1478302001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 80, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19991594105958937, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.017634215485304593, + "grad_norm": 1.5727171897888184, + "kept_token_count": 654.5671875, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.8082, + "mgd_cos_sim": 0.19180826246738433, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.8081917345523835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 23.33884201049805, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8639678955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 90, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19966457784175873, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.01628115987405181, + "grad_norm": 1.5600043535232544, + "kept_token_count": 654.9703125, + "learning_rate": 1.98e-06, + "loss": 0.7681, + "mgd_cos_sim": 0.23194781839847564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.7680521845817566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 22.77388725280762, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.51283874511716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 100, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.20058237612247468, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.01917586401104927, + "grad_norm": 1.5338762998580933, + "kept_token_count": 653.9953125, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.7198, + "mgd_cos_sim": 0.2802122086286545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.7197877883911132, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 21.880190658569337, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4451873779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 110, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19960569888353347, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.01624835543334484, + "grad_norm": 1.4215947389602661, + "kept_token_count": 654.8578125, + "learning_rate": 2.38e-06, + "loss": 0.6619, + "mgd_cos_sim": 0.3380711317062378, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.6619288504123688, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 21.44614429473877, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8155883789062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 120, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20127865374088288, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.017582985013723372, + "grad_norm": 1.2588701248168945, + "kept_token_count": 653.278125, + "learning_rate": 2.58e-06, + "loss": 0.6036, + "mgd_cos_sim": 0.3963611990213394, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.6036387920379639, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 21.203883552551268, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0872863769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 130, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.20008667111396788, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.0188486997038126, + "grad_norm": 1.1285250186920166, + "kept_token_count": 654.609375, + "learning_rate": 2.78e-06, + "loss": 0.5448, + "mgd_cos_sim": 0.4552246928215027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.5447753012180329, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 21.50022163391113, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.29688415527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1993465319275856, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.01937663508579135, + "grad_norm": 1.0208072662353516, + "kept_token_count": 655.0921875, + "learning_rate": 2.98e-06, + "loss": 0.4949, + "mgd_cos_sim": 0.5051402777433396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.49485973119735716, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 22.183237648010255, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6585754394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 150, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19965860098600388, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.018002827558666466, + "grad_norm": 0.9323387742042542, + "kept_token_count": 654.95625, + "learning_rate": 3.18e-06, + "loss": 0.4468, + "mgd_cos_sim": 0.5532383620738983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.4467616558074951, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 23.015914154052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.85552368164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 160, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19919416755437852, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.017104405257850886, + "grad_norm": 0.8680227994918823, + "kept_token_count": 655.040625, + "learning_rate": 3.38e-06, + "loss": 0.4111, + "mgd_cos_sim": 0.5888814628124237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.41111853420734407, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 24.120131301879884, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5504455566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.1996718928217888, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.01593392910435796, + "grad_norm": 0.8413816690444946, + "kept_token_count": 655.0125, + "learning_rate": 3.58e-06, + "loss": 0.3755, + "mgd_cos_sim": 0.6244669437408448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.37553304731845855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 25.31215057373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.51740112304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 180, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.1996667891740799, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.016998144704848527, + "grad_norm": 0.796542227268219, + "kept_token_count": 655.0484375, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.3488, + "mgd_cos_sim": 0.6511636257171631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.34883637726306915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 26.187657356262207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5902099609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 190, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.20054907500743865, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.018794136866927148, + "grad_norm": 0.7844157218933105, + "kept_token_count": 654.1140625, + "learning_rate": 3.98e-06, + "loss": 0.3195, + "mgd_cos_sim": 0.6804613590240478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.31953864097595214, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 26.88565673828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6094940185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 200, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19938467890024186, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.020166187454015015, + "grad_norm": 0.7253143787384033, + "kept_token_count": 655.3859375, + "learning_rate": 4.18e-06, + "loss": 0.3004, + "mgd_cos_sim": 0.6996039927005768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.300396004319191, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.422644233703615, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5216003417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 210, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.199477855861187, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.01879007602110505, + "grad_norm": 0.7031867504119873, + "kept_token_count": 654.7015625, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.283, + "mgd_cos_sim": 0.7169580459594727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.2830419421195984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.772640991210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3504913330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 220, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.20023885518312454, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.02316160425543785, + "grad_norm": 0.6660106778144836, + "kept_token_count": 654.1375, + "learning_rate": 4.58e-06, + "loss": 0.2659, + "mgd_cos_sim": 0.7340972602367402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.2659027293324471, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.67985553741455, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.73993225097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 230, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.19943009316921234, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.02598432581871748, + "grad_norm": 0.6276301741600037, + "kept_token_count": 654.840625, + "learning_rate": 4.78e-06, + "loss": 0.2468, + "mgd_cos_sim": 0.7531943500041962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.24680565893650055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.702269744873046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8728759765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 240, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.20083319544792175, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.022919824346899986, + "grad_norm": 0.6099145412445068, + "kept_token_count": 654.1609375, + "learning_rate": 4.980000000000001e-06, + "loss": 0.2294, + "mgd_cos_sim": 0.7705694496631622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.22943055033683776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.641964530944826, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6803009033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 250, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.20000607371330262, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.023281680420041084, + "grad_norm": 0.5950484871864319, + "kept_token_count": 654.628125, + "learning_rate": 5.18e-06, + "loss": 0.2138, + "mgd_cos_sim": 0.786239755153656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.21376024335622787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.748721313476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3746368408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 260, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2004251465201378, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.026643300615251064, + "grad_norm": 0.5119501948356628, + "kept_token_count": 654.1453125, + "learning_rate": 5.38e-06, + "loss": 0.1972, + "mgd_cos_sim": 0.8028217494487763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.19717824310064316, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 27.910375022888182, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 270, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20051535069942475, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.02522715665400028, + "grad_norm": 0.5241382122039795, + "kept_token_count": 654.0734375, + "learning_rate": 5.58e-06, + "loss": 0.1868, + "mgd_cos_sim": 0.8131818830966949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1868181124329567, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.058662796020506, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3865509033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 280, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.19981991946697236, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.026847702264785767, + "grad_norm": 0.4418611526489258, + "kept_token_count": 654.64375, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.1693, + "mgd_cos_sim": 0.8307357370853424, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.16926427632570268, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.199527931213378, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.22491760253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 290, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19951998442411423, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.033483318611979486, + "grad_norm": 0.4024133086204529, + "kept_token_count": 654.7828125, + "learning_rate": 5.98e-06, + "loss": 0.1664, + "mgd_cos_sim": 0.833588707447052, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.16641127914190293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.3351692199707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0365447998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 300, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.2006666287779808, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.034327675215899946, + "grad_norm": 0.3838469684123993, + "kept_token_count": 654.1296875, + "learning_rate": 6.18e-06, + "loss": 0.1581, + "mgd_cos_sim": 0.8419468522071838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.15805312842130662, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.45145950317383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4496612548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19948843866586685, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.0288911871612072, + "grad_norm": 0.33290398120880127, + "kept_token_count": 654.9890625, + "learning_rate": 6.38e-06, + "loss": 0.1479, + "mgd_cos_sim": 0.8520561099052429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1479438930749893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.753381729125977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5890838623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 320, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20026820302009582, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.0373080613091588, + "grad_norm": 0.31135499477386475, + "kept_token_count": 654.2765625, + "learning_rate": 6.58e-06, + "loss": 0.1426, + "mgd_cos_sim": 0.8574293851852417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.142570623755455, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 28.922828102111815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1398162841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 330, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.1996892511844635, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.04410234205424786, + "grad_norm": 0.309424489736557, + "kept_token_count": 654.9046875, + "learning_rate": 6.78e-06, + "loss": 0.138, + "mgd_cos_sim": 0.8620357394218445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.13796425312757493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 29.27288188934326, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05101623535154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 340, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.200026074051857, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.03970141187310219, + "grad_norm": 0.34514763951301575, + "kept_token_count": 654.35625, + "learning_rate": 6.98e-06, + "loss": 0.1317, + "mgd_cos_sim": 0.8682890832424164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1317109189927578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 29.671848678588866, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3957183837891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 350, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.1983961582183838, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.044667305797338484, + "grad_norm": 0.2091762125492096, + "kept_token_count": 656.13125, + "learning_rate": 7.180000000000001e-06, + "loss": 0.1325, + "mgd_cos_sim": 0.8675284028053284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.13247158676385878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 29.96899356842041, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.02057495117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 360, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.2004110500216484, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.042445831932127476, + "grad_norm": 0.23838739097118378, + "kept_token_count": 654.4125, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.1295, + "mgd_cos_sim": 0.8704963445663452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12950363904237747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 30.347439765930176, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.05467834472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 370, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.1997595578432083, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.04481091052293777, + "grad_norm": 0.21298059821128845, + "kept_token_count": 654.759375, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.125, + "mgd_cos_sim": 0.8750308752059937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12496913596987724, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 30.77487506866455, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.73610534667966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 380, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20006652921438217, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.046141424030065534, + "grad_norm": 0.20614369213581085, + "kept_token_count": 654.41875, + "learning_rate": 7.780000000000002e-06, + "loss": 0.1164, + "mgd_cos_sim": 0.88360755443573, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11639244705438614, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 31.120339393615723, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5226165771484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 390, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.20090126246213913, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.04751823507249355, + "grad_norm": 0.17637933790683746, + "kept_token_count": 653.81875, + "learning_rate": 7.98e-06, + "loss": 0.1228, + "mgd_cos_sim": 0.8772078931331635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12279209569096565, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 31.634296798706053, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7953857421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 400, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20002700686454772, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.0516414787620306, + "grad_norm": 0.15428653359413147, + "kept_token_count": 654.403125, + "learning_rate": 8.18e-06, + "loss": 0.1174, + "mgd_cos_sim": 0.8826488375663757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11735116243362427, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 32.04456062316895, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0543731689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 410, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.1997506231069565, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.05272260010242462, + "grad_norm": 0.2770257294178009, + "kept_token_count": 654.8328125, + "learning_rate": 8.38e-06, + "loss": 0.1203, + "mgd_cos_sim": 0.8796727299690247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12032726407051086, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 32.415503311157224, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7174011230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 420, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20042356550693513, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.06010093912482262, + "grad_norm": 0.261120468378067, + "kept_token_count": 654.184375, + "learning_rate": 8.58e-06, + "loss": 0.1238, + "mgd_cos_sim": 0.8761562645435333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12384373769164085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 32.86924858093262, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.36285705566405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.1999556303024292, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.05726204514503479, + "grad_norm": 0.2168649584054947, + "kept_token_count": 654.8828125, + "learning_rate": 8.78e-06, + "loss": 0.1247, + "mgd_cos_sim": 0.8753358542919158, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12466415166854858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 33.28693046569824, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.44122314453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 440, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.1989833027124405, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.054698488861322406, + "grad_norm": 0.19059276580810547, + "kept_token_count": 655.3625, + "learning_rate": 8.98e-06, + "loss": 0.1156, + "mgd_cos_sim": 0.8843732535839081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1156267449259758, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 33.72381782531738, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.67381896972654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.1995469942688942, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.057995780184864995, + "grad_norm": 0.1914421021938324, + "kept_token_count": 654.98125, + "learning_rate": 9.18e-06, + "loss": 0.1184, + "mgd_cos_sim": 0.881557846069336, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11844213977456093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 34.07507743835449, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1535919189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 460, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20002176612615585, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.06419377624988556, + "grad_norm": 0.13910160958766937, + "kept_token_count": 654.65625, + "learning_rate": 9.38e-06, + "loss": 0.1237, + "mgd_cos_sim": 0.8763318717479706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12366811484098435, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 34.58454818725586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.1310272216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 470, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19984840750694274, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.061857306957244874, + "grad_norm": 0.1808994859457016, + "kept_token_count": 654.8609375, + "learning_rate": 9.580000000000001e-06, + "loss": 0.1134, + "mgd_cos_sim": 0.8866123378276825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1133876658976078, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 35.03890762329102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0614410400391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 480, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.1991987034678459, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.0685441043227911, + "grad_norm": 0.16103017330169678, + "kept_token_count": 655.3796875, + "learning_rate": 9.780000000000001e-06, + "loss": 0.116, + "mgd_cos_sim": 0.8839855670928956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11601443290710449, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 35.587548446655276, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.04591369628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 490, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.200205397605896, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.06606069020926952, + "grad_norm": 0.23109309375286102, + "kept_token_count": 654.209375, + "learning_rate": 9.980000000000001e-06, + "loss": 0.1129, + "mgd_cos_sim": 0.8870607316493988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11293926909565925, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 36.18183860778809, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1562774658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 500, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.1998801276087761, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.0642350424081087, + "grad_norm": 0.17936386168003082, + "kept_token_count": 654.653125, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.1102, + "mgd_cos_sim": 0.8898132801055908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11018671989440917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 36.77784957885742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0479736328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 510, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.19973150491714478, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.07595390751957894, + "grad_norm": 0.16317467391490936, + "kept_token_count": 654.9171875, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.1209, + "mgd_cos_sim": 0.879134327173233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.12086567059159278, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 37.332058334350585, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.23060302734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 520, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20116025060415268, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.07400693707168102, + "grad_norm": 0.1631965935230255, + "kept_token_count": 653.615625, + "learning_rate": 1.058e-05, + "loss": 0.1136, + "mgd_cos_sim": 0.8864198684692383, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11358013227581978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 37.85455169677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9484069824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 530, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2005286008119583, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.08299172669649124, + "grad_norm": 0.24589046835899353, + "kept_token_count": 654.1046875, + "learning_rate": 1.078e-05, + "loss": 0.1141, + "mgd_cos_sim": 0.8858941316604614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11410587579011917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 38.33886375427246, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9523406982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 540, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20003687590360641, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.07600862421095371, + "grad_norm": 0.16269434988498688, + "kept_token_count": 654.809375, + "learning_rate": 1.098e-05, + "loss": 0.1149, + "mgd_cos_sim": 0.8851048588752747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11489512622356415, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 38.9089958190918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.68186645507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 550, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.20084057748317719, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.07854113355278969, + "grad_norm": 0.23208187520503998, + "kept_token_count": 653.7671875, + "learning_rate": 1.118e-05, + "loss": 0.1111, + "mgd_cos_sim": 0.8889387667179107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.11106122434139251, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 39.415116500854495, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9859893798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 560, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.19993816018104554, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.08471892960369587, + "grad_norm": 0.1791197508573532, + "kept_token_count": 654.609375, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.1137, + "mgd_cos_sim": 0.8862728357315064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.113727168738842, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 39.68782081604004, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.49118957519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 570, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.1995413288474083, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.08320528045296668, + "grad_norm": 0.14353254437446594, + "kept_token_count": 654.8046875, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.1096, + "mgd_cos_sim": 0.8903883159160614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10961168333888054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 40.13898544311523, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2016906738281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 580, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.1998854786157608, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.08431864753365517, + "grad_norm": 0.1250118464231491, + "kept_token_count": 654.69375, + "learning_rate": 1.178e-05, + "loss": 0.1082, + "mgd_cos_sim": 0.8917904198169708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10820958986878396, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 40.522774124145506, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6104766845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 590, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19943167567253112, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.08515657335519791, + "grad_norm": 0.187725231051445, + "kept_token_count": 655.15, + "learning_rate": 1.198e-05, + "loss": 0.1088, + "mgd_cos_sim": 0.8912371039390564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10876289680600167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 40.86361999511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9385955810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 600, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19998343288898468, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.08020749911665917, + "grad_norm": 0.14741411805152893, + "kept_token_count": 654.5171875, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.1018, + "mgd_cos_sim": 0.8982001602649688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10179984197020531, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 41.315900802612305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6513397216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 610, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.199781434237957, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.08542994633316994, + "grad_norm": 0.19807900488376617, + "kept_token_count": 654.6203125, + "learning_rate": 1.238e-05, + "loss": 0.1063, + "mgd_cos_sim": 0.8937248408794403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10627517625689506, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 41.78718528747559, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3738311767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 620, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.1999923586845398, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.08830841109156609, + "grad_norm": 0.1642083376646042, + "kept_token_count": 654.715625, + "learning_rate": 1.258e-05, + "loss": 0.1038, + "mgd_cos_sim": 0.8961813688278198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10381861925125122, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 42.136850357055664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1742767333984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 630, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.19967905133962632, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.08874529749155044, + "grad_norm": 0.15096250176429749, + "kept_token_count": 654.809375, + "learning_rate": 1.278e-05, + "loss": 0.1048, + "mgd_cos_sim": 0.8952026128768921, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1047973908483982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 42.31564445495606, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.34229431152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 640, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.20083203613758088, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.09379860013723373, + "grad_norm": 0.15910698473453522, + "kept_token_count": 653.8140625, + "learning_rate": 1.298e-05, + "loss": 0.1066, + "mgd_cos_sim": 0.8934344172477722, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10656559318304062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 42.4168342590332, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1734130859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 650, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20061006546020507, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.09596546739339828, + "grad_norm": 0.18959824740886688, + "kept_token_count": 654.06875, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.1067, + "mgd_cos_sim": 0.8933461666107178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10665384754538536, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 42.708787536621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.39906616210936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 660, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.20031684786081314, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.08304863721132279, + "grad_norm": 0.18625687062740326, + "kept_token_count": 654.45, + "learning_rate": 1.338e-05, + "loss": 0.1015, + "mgd_cos_sim": 0.898505973815918, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10149401128292083, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 42.921226501464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8258850097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 670, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.20093102008104324, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.08475578874349594, + "grad_norm": 0.22609388828277588, + "kept_token_count": 653.8015625, + "learning_rate": 1.358e-05, + "loss": 0.1026, + "mgd_cos_sim": 0.8973930776119232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1026069276034832, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 43.31545181274414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6218292236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 680, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.1998279109597206, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.09206265956163406, + "grad_norm": 0.21757954359054565, + "kept_token_count": 654.83125, + "learning_rate": 1.378e-05, + "loss": 0.1041, + "mgd_cos_sim": 0.8958615720272064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10413841307163238, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 43.664744186401364, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.51904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 690, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19977016001939774, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.09402164593338966, + "grad_norm": 0.21515923738479614, + "kept_token_count": 655.0015625, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.1068, + "mgd_cos_sim": 0.8932005107402802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.1067995011806488, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 44.00968208312988, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2043090820313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 700, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19972479045391084, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.09176048710942268, + "grad_norm": 0.17563050985336304, + "kept_token_count": 655.103125, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.1013, + "mgd_cos_sim": 0.8986981570720672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10130185037851333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 44.480668258666995, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7368408203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 710, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.19917880594730378, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.08685743808746338, + "grad_norm": 0.22104427218437195, + "kept_token_count": 655.0859375, + "learning_rate": 1.438e-05, + "loss": 0.1005, + "mgd_cos_sim": 0.8994569301605224, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10054307878017425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 44.639313507080075, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.85260620117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 720, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.19956449717283248, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.08649691045284272, + "grad_norm": 0.15156179666519165, + "kept_token_count": 655.0375, + "learning_rate": 1.458e-05, + "loss": 0.0993, + "mgd_cos_sim": 0.9006953537464142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09930464401841163, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 44.96775932312012, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9242645263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 730, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19906608909368514, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.09368709549307823, + "grad_norm": 0.17833814024925232, + "kept_token_count": 655.325, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.1018, + "mgd_cos_sim": 0.8981734871864319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10182652249932289, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 44.94701232910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1270233154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 740, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20015522539615632, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.08948827758431435, + "grad_norm": 0.17677801847457886, + "kept_token_count": 654.590625, + "learning_rate": 1.498e-05, + "loss": 0.1001, + "mgd_cos_sim": 0.8999351799488068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10006482005119324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 45.31419448852539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4358795166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 750, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20055359303951265, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.08976704031229019, + "grad_norm": 0.1405053585767746, + "kept_token_count": 654.0515625, + "learning_rate": 1.518e-05, + "loss": 0.097, + "mgd_cos_sim": 0.9029895424842834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09701045453548432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 45.5617130279541, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6730438232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 760, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.20074717700481415, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.0965663067996502, + "grad_norm": 0.17872396111488342, + "kept_token_count": 653.83125, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.1062, + "mgd_cos_sim": 0.8937992036342621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.10620079562067986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 45.56846733093262, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.43540344238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 770, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.19997320622205733, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.08249439597129822, + "grad_norm": 0.22452615201473236, + "kept_token_count": 654.484375, + "learning_rate": 1.558e-05, + "loss": 0.0993, + "mgd_cos_sim": 0.9007054209709168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09929458051919937, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 45.92748489379883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2691680908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 780, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.20086227357387543, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.09115035161375999, + "grad_norm": 0.1695842444896698, + "kept_token_count": 653.540625, + "learning_rate": 1.578e-05, + "loss": 0.0985, + "mgd_cos_sim": 0.9015036225318909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09849637597799302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.129059600830075, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1736145019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 790, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.19955798238515854, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.08816298246383666, + "grad_norm": 0.1517307013273239, + "kept_token_count": 654.7296875, + "learning_rate": 1.598e-05, + "loss": 0.0985, + "mgd_cos_sim": 0.9014753878116608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09852460846304893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.42261276245117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.34108581542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 800, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.19970413595438002, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.08197291158139705, + "grad_norm": 0.20400656759738922, + "kept_token_count": 654.7734375, + "learning_rate": 1.618e-05, + "loss": 0.0982, + "mgd_cos_sim": 0.9017996370792389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09820037186145783, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.5639404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6623504638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 810, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19977138787508011, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.08969309628009796, + "grad_norm": 0.18586844205856323, + "kept_token_count": 654.64375, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0992, + "mgd_cos_sim": 0.9008011877536773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0991988092660904, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.645346069335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3008331298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 820, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20039928704500198, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.08372624367475509, + "grad_norm": 0.16513654589653015, + "kept_token_count": 654.040625, + "learning_rate": 1.658e-05, + "loss": 0.0949, + "mgd_cos_sim": 0.9051079869270324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09489200413227081, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.92005157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7674530029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 830, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.20093437433242797, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.08441628962755203, + "grad_norm": 0.19112516939640045, + "kept_token_count": 653.8796875, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0982, + "mgd_cos_sim": 0.9017789244651795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09822105467319489, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.86205062866211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8290069580078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 840, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.19983178228139878, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.081609296053648, + "grad_norm": 0.2285463809967041, + "kept_token_count": 654.6890625, + "learning_rate": 1.698e-05, + "loss": 0.097, + "mgd_cos_sim": 0.9029880225658417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0970119632780552, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 46.9533016204834, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2893341064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 850, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19952128678560258, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.07680138647556305, + "grad_norm": 0.25133487582206726, + "kept_token_count": 655.2328125, + "learning_rate": 1.718e-05, + "loss": 0.0886, + "mgd_cos_sim": 0.9114183604717254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08858163207769394, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 47.2642707824707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.78147888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 860, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.19994822144508362, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.08031991943717003, + "grad_norm": 0.19541457295417786, + "kept_token_count": 654.38125, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.092, + "mgd_cos_sim": 0.9079780459403992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09202195033431053, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 47.42965431213379, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6517822265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 870, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.20111705362796783, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.08242396414279937, + "grad_norm": 0.18830661475658417, + "kept_token_count": 653.9265625, + "learning_rate": 1.758e-05, + "loss": 0.0938, + "mgd_cos_sim": 0.9061697542667388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09383025541901588, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 47.83159484863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.59742431640626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 880, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.2007051944732666, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.07890086024999618, + "grad_norm": 0.25727272033691406, + "kept_token_count": 653.8421875, + "learning_rate": 1.778e-05, + "loss": 0.0902, + "mgd_cos_sim": 0.9098159909248352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.09018399640917778, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 48.1399642944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.02923278808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 890, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.19927389323711395, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.07935443557798863, + "grad_norm": 0.23035138845443726, + "kept_token_count": 655.1234375, + "learning_rate": 1.798e-05, + "loss": 0.0887, + "mgd_cos_sim": 0.9113218724727631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08867813050746917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 48.58092613220215, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.042626953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 900, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.19968797862529755, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.07663023993372917, + "grad_norm": 0.2082872837781906, + "kept_token_count": 654.7140625, + "learning_rate": 1.818e-05, + "loss": 0.089, + "mgd_cos_sim": 0.9110042035579682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08899579048156739, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 49.10774688720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3035186767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 910, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.19962272346019744, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.08099409714341163, + "grad_norm": 0.17440840601921082, + "kept_token_count": 655.0125, + "learning_rate": 1.838e-05, + "loss": 0.0853, + "mgd_cos_sim": 0.9147037446498871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08529625311493874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.257886505126955, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0355712890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 920, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.20024116486310958, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.08698826804757118, + "grad_norm": 0.23135216534137726, + "kept_token_count": 654.5390625, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0883, + "mgd_cos_sim": 0.9117366015911103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08826339840888978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.59366798400879, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0800415039063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 930, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20042355507612228, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.07845558486878872, + "grad_norm": 0.1661590337753296, + "kept_token_count": 654.2734375, + "learning_rate": 1.878e-05, + "loss": 0.0884, + "mgd_cos_sim": 0.9115680813789367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08843192383646965, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.725569534301755, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.90611572265624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 940, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20010186731815338, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.07301427237689495, + "grad_norm": 0.17889006435871124, + "kept_token_count": 654.8546875, + "learning_rate": 1.898e-05, + "loss": 0.0848, + "mgd_cos_sim": 0.9151930749416352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08480691462755204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.794837951660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6521057128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 950, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.2008995682001114, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.07861691154539585, + "grad_norm": 0.16612908244132996, + "kept_token_count": 654.103125, + "learning_rate": 1.918e-05, + "loss": 0.0818, + "mgd_cos_sim": 0.9182069957256317, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08179300799965858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 50.896779251098636, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4801086425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 960, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.19968316853046417, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.08483279421925545, + "grad_norm": 0.20777469873428345, + "kept_token_count": 654.8765625, + "learning_rate": 1.938e-05, + "loss": 0.0858, + "mgd_cos_sim": 0.9141830086708069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08581701070070266, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 51.64723892211914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4647155761719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2000775694847107, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.07378560304641724, + "grad_norm": 0.23898504674434662, + "kept_token_count": 654.3953125, + "learning_rate": 1.958e-05, + "loss": 0.0819, + "mgd_cos_sim": 0.9181330919265747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08186690360307694, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.17413063049317, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3227081298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 980, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.2011079892516136, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.08779957816004753, + "grad_norm": 0.22059515118598938, + "kept_token_count": 653.9890625, + "learning_rate": 1.978e-05, + "loss": 0.0891, + "mgd_cos_sim": 0.9109064280986786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08909357935190201, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.62125358581543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.99884033203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 990, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.19998737722635268, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.08590830117464066, + "grad_norm": 0.16893616318702698, + "kept_token_count": 654.684375, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0859, + "mgd_cos_sim": 0.9140590369701386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08594096601009368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 52.92889633178711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.60346069335935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1000, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19992413967847825, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.08026496581733226, + "grad_norm": 0.1659822016954422, + "kept_token_count": 654.6875, + "learning_rate": 2.018e-05, + "loss": 0.0813, + "mgd_cos_sim": 0.91872478723526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.08127520605921745, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 53.79764251708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2653289794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1010, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20031221359968185, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.08366950266063214, + "grad_norm": 0.18378648161888123, + "kept_token_count": 654.3125, + "learning_rate": 2.038e-05, + "loss": 0.079, + "mgd_cos_sim": 0.9209749817848205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07902501225471496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.14103965759277, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.88336181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19974431842565538, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.07781486809253693, + "grad_norm": 0.16706639528274536, + "kept_token_count": 654.9671875, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0747, + "mgd_cos_sim": 0.9253381073474884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07466189563274384, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.95926094055176, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9408447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1030, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.19992035180330275, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.08000589869916439, + "grad_norm": 0.17674526572227478, + "kept_token_count": 654.7890625, + "learning_rate": 2.078e-05, + "loss": 0.0788, + "mgd_cos_sim": 0.9212157905101777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07878421023488044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 54.900977325439456, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2892364501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1040, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19961424618959428, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.07716933414340019, + "grad_norm": 0.24737855792045593, + "kept_token_count": 655.265625, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0775, + "mgd_cos_sim": 0.9224752426147461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07752476818859577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.19161491394043, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2528015136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1050, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.20022748410701752, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.07462317608296871, + "grad_norm": 0.15718092024326324, + "kept_token_count": 654.321875, + "learning_rate": 2.118e-05, + "loss": 0.0729, + "mgd_cos_sim": 0.9270883560180664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07291165962815285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 55.754848861694335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.60332641601565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1060, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20073560923337935, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.08577392101287842, + "grad_norm": 0.14486286044120789, + "kept_token_count": 654.1578125, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0727, + "mgd_cos_sim": 0.9273142337799072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07268575578927994, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 56.99715919494629, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8540405273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1070, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.20039655566215514, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.08539300300180912, + "grad_norm": 0.1775049865245819, + "kept_token_count": 654.346875, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0752, + "mgd_cos_sim": 0.9248351454734802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07516485303640366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.21102104187012, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3746002197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1080, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.1993764042854309, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.0837398737668991, + "grad_norm": 0.16953900456428528, + "kept_token_count": 655.1015625, + "learning_rate": 2.178e-05, + "loss": 0.0757, + "mgd_cos_sim": 0.9243146419525147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07568536251783371, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.33056411743164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.45558166503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1090, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19973916858434676, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.0811839945614338, + "grad_norm": 0.2064359486103058, + "kept_token_count": 654.784375, + "learning_rate": 2.198e-05, + "loss": 0.0741, + "mgd_cos_sim": 0.9258521020412445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07414789497852325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 57.933203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2779205322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1100, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.19993724972009658, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.08444629311561584, + "grad_norm": 0.16536575555801392, + "kept_token_count": 654.715625, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0704, + "mgd_cos_sim": 0.929638797044754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07036119550466538, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 58.77368583679199, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2042266845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1110, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19985413402318955, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.0819057036191225, + "grad_norm": 0.19151857495307922, + "kept_token_count": 654.853125, + "learning_rate": 2.238e-05, + "loss": 0.0704, + "mgd_cos_sim": 0.9295702517032624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07042975164949894, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 59.557569885253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41808166503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1120, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19922174662351608, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.09098964817821979, + "grad_norm": 0.23900113999843597, + "kept_token_count": 655.3390625, + "learning_rate": 2.258e-05, + "loss": 0.0708, + "mgd_cos_sim": 0.9292065739631653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07079343758523464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.70598945617676, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.59093322753904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1130, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.19967683106660844, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.09556545317173004, + "grad_norm": 0.17968977987766266, + "kept_token_count": 655.078125, + "learning_rate": 2.278e-05, + "loss": 0.0722, + "mgd_cos_sim": 0.9277899384498596, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.07221006080508233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 60.74554977416992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0485534667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1140, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19897326231002807, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.08405668064951896, + "grad_norm": 0.23731860518455505, + "kept_token_count": 655.3375, + "learning_rate": 2.298e-05, + "loss": 0.0685, + "mgd_cos_sim": 0.931450343132019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06854963600635529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 61.112618637084964, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3173004150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1150, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19975287318229676, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.08631790615618229, + "grad_norm": 0.24645933508872986, + "kept_token_count": 654.8515625, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0674, + "mgd_cos_sim": 0.932644146680832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06735585778951644, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.10214424133301, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5011688232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1160, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.1994330883026123, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.09579779356718063, + "grad_norm": 0.2572055459022522, + "kept_token_count": 654.8171875, + "learning_rate": 2.338e-05, + "loss": 0.0667, + "mgd_cos_sim": 0.9332628071308136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06673719622194767, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.87805442810058, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.780859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1170, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.1999084547162056, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.09708105027675629, + "grad_norm": 0.24858912825584412, + "kept_token_count": 654.8171875, + "learning_rate": 2.358e-05, + "loss": 0.0684, + "mgd_cos_sim": 0.9316193163394928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06838067770004272, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 62.94683151245117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.55933837890626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1180, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.19970900565385818, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.0872678093612194, + "grad_norm": 0.23745191097259521, + "kept_token_count": 654.91875, + "learning_rate": 2.378e-05, + "loss": 0.0651, + "mgd_cos_sim": 0.9349134266376495, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06508656889200211, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 63.509588623046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.83677673339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1190, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.19916920959949494, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.09585636183619499, + "grad_norm": 0.353331595659256, + "kept_token_count": 655.4640625, + "learning_rate": 2.398e-05, + "loss": 0.064, + "mgd_cos_sim": 0.9360343933105468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06396560780704022, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.14280776977539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7078918457031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1200, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.20027067214250566, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.08870217502117157, + "grad_norm": 0.23108190298080444, + "kept_token_count": 654.1109375, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0625, + "mgd_cos_sim": 0.9375481069087982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.062451893091201784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.77237548828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.28350524902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1210, + "valid_token_count": 817.9078125 + }, + { + "actual_token_mask_ratio": 0.20008421838283538, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.09122511968016625, + "grad_norm": 0.18050792813301086, + "kept_token_count": 654.5296875, + "learning_rate": 2.438e-05, + "loss": 0.0607, + "mgd_cos_sim": 0.9393199443817138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.060680067539215087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 64.80373497009278, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7229248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1220, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1992115333676338, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.09840098172426223, + "grad_norm": 0.1956564038991928, + "kept_token_count": 654.8453125, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0611, + "mgd_cos_sim": 0.9388892650604248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06111072897911072, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 66.60762786865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9437255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1230, + "valid_token_count": 817.75625 + }, + { + "actual_token_mask_ratio": 0.19969327002763748, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.09601365849375725, + "grad_norm": 0.18559914827346802, + "kept_token_count": 654.8625, + "learning_rate": 2.478e-05, + "loss": 0.0614, + "mgd_cos_sim": 0.9386380434036254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.06136196181178093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 67.39400177001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1882781982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1240, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19955934286117555, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.09773015603423119, + "grad_norm": 0.267362117767334, + "kept_token_count": 654.9015625, + "learning_rate": 2.498e-05, + "loss": 0.0567, + "mgd_cos_sim": 0.9433341383934021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.056665873154997826, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 68.65859451293946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83184509277345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1250, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.20028460174798965, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.09626142084598541, + "grad_norm": 0.19638372957706451, + "kept_token_count": 654.4953125, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0536, + "mgd_cos_sim": 0.9463570594787598, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.053642938286066054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.5272834777832, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.29600219726564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1260, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20030070692300797, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.09270666465163231, + "grad_norm": 0.17225976288318634, + "kept_token_count": 654.3359375, + "learning_rate": 2.538e-05, + "loss": 0.0562, + "mgd_cos_sim": 0.9438360214233399, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.056163974851369855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 69.61979064941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1270, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20012175440788268, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.09293120577931405, + "grad_norm": 0.16232508420944214, + "kept_token_count": 654.528125, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.051, + "mgd_cos_sim": 0.9489751040935517, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.05102487541735172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.3502426147461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0589996337891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1280, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20039659440517427, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.10104202777147293, + "grad_norm": 0.16796085238456726, + "kept_token_count": 654.0, + "learning_rate": 2.578e-05, + "loss": 0.0552, + "mgd_cos_sim": 0.9447907567024231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.05520922802388668, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 70.2922981262207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7018188476562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1290, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.19958732277154922, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.09992209002375603, + "grad_norm": 0.2653164267539978, + "kept_token_count": 654.8171875, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0527, + "mgd_cos_sim": 0.9472730338573456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.05272697359323501, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.0191764831543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8940124511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1300, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.20052043348550797, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.09471441432833672, + "grad_norm": 0.19453614950180054, + "kept_token_count": 654.2515625, + "learning_rate": 2.618e-05, + "loss": 0.0526, + "mgd_cos_sim": 0.9473569691181183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.052643037587404254, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.26387329101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9543762207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.20006825774908066, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.0820510134100914, + "grad_norm": 0.26405876874923706, + "kept_token_count": 654.5484375, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0488, + "mgd_cos_sim": 0.9512086033821106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.048791400715708734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 71.85520858764649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3080627441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1320, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19998738616704942, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.10083143785595894, + "grad_norm": 0.19039319455623627, + "kept_token_count": 654.7625, + "learning_rate": 2.658e-05, + "loss": 0.0524, + "mgd_cos_sim": 0.9475605845451355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.05243940912187099, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.15868225097657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5260894775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1330, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.1995933771133423, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.09149783253669738, + "grad_norm": 0.1925138235092163, + "kept_token_count": 655.203125, + "learning_rate": 2.678e-05, + "loss": 0.0481, + "mgd_cos_sim": 0.9518611133098602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04813888296484947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 72.80649948120117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5879211425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1340, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.19966234415769576, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.10200502052903175, + "grad_norm": 0.2581107020378113, + "kept_token_count": 654.8984375, + "learning_rate": 2.698e-05, + "loss": 0.0492, + "mgd_cos_sim": 0.9508018672466279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04919813238084316, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 73.36094360351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7269561767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1350, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.1997348964214325, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.1010720632970333, + "grad_norm": 0.22686561942100525, + "kept_token_count": 654.7921875, + "learning_rate": 2.718e-05, + "loss": 0.0494, + "mgd_cos_sim": 0.9506396174430847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.049360374733805654, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 75.96223373413086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.37346801757815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1360, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.1998886689543724, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.10263365432620049, + "grad_norm": 0.15959042310714722, + "kept_token_count": 654.8203125, + "learning_rate": 2.738e-05, + "loss": 0.05, + "mgd_cos_sim": 0.949978131055832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.05002186521887779, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.57123107910157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5013031005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1370, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20053405016660691, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.09530559927225113, + "grad_norm": 0.1980428546667099, + "kept_token_count": 654.365625, + "learning_rate": 2.758e-05, + "loss": 0.0453, + "mgd_cos_sim": 0.954701566696167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04529844261705875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 76.69036026000977, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1434387207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1380, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.1992841586470604, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.09952058419585227, + "grad_norm": 0.17337489128112793, + "kept_token_count": 655.0890625, + "learning_rate": 2.778e-05, + "loss": 0.0456, + "mgd_cos_sim": 0.954370242357254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04562976695597172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 77.94293746948242, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47321166992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1390, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19993306547403336, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.10148442760109902, + "grad_norm": 0.13440562784671783, + "kept_token_count": 654.8484375, + "learning_rate": 2.798e-05, + "loss": 0.0462, + "mgd_cos_sim": 0.9538329422473908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04616704881191254, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 79.09429016113282, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7548004150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1400, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2001159131526947, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.10161614641547204, + "grad_norm": 0.18880584836006165, + "kept_token_count": 654.578125, + "learning_rate": 2.818e-05, + "loss": 0.0441, + "mgd_cos_sim": 0.9559123277664184, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04408766962587833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.95042037963867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5883483886719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1410, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20043645054101944, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.10913017839193344, + "grad_norm": 0.2460772544145584, + "kept_token_count": 654.1828125, + "learning_rate": 2.838e-05, + "loss": 0.0446, + "mgd_cos_sim": 0.9554270923137664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04457290433347225, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 80.93573379516602, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94720153808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1420, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20019663274288177, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.10598045513033867, + "grad_norm": 0.15617409348487854, + "kept_token_count": 654.3671875, + "learning_rate": 2.858e-05, + "loss": 0.0459, + "mgd_cos_sim": 0.9540988385677338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.045901141688227655, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 81.27484970092773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2399566650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20087729543447494, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.10663834810256959, + "grad_norm": 0.20398567616939545, + "kept_token_count": 653.8140625, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0441, + "mgd_cos_sim": 0.9559085905551911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04409141205251217, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 82.18357162475586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.55494384765626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1440, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20009985715150833, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.11099661812186241, + "grad_norm": 0.1905137300491333, + "kept_token_count": 654.58125, + "learning_rate": 2.898e-05, + "loss": 0.0422, + "mgd_cos_sim": 0.9578351438045501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04216484278440476, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.44282760620118, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0570404052734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1450, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.199893981218338, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.11767789721488953, + "grad_norm": 0.1631869226694107, + "kept_token_count": 654.56875, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0441, + "mgd_cos_sim": 0.955904757976532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.044095227122306825, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 83.45075454711915, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4552795410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1460, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.20044448226690292, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.11122365221381188, + "grad_norm": 0.18531066179275513, + "kept_token_count": 654.2890625, + "learning_rate": 2.938e-05, + "loss": 0.0432, + "mgd_cos_sim": 0.956827187538147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0431728184223175, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.87523880004883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98854064941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1470, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2003392830491066, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.10443519502878189, + "grad_norm": 0.15246549248695374, + "kept_token_count": 654.4953125, + "learning_rate": 2.958e-05, + "loss": 0.0422, + "mgd_cos_sim": 0.9578020632266998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.042197947204113004, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.91030349731446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2269989013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1480, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.2005782350897789, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.10621093064546586, + "grad_norm": 0.2335565686225891, + "kept_token_count": 654.10625, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0415, + "mgd_cos_sim": 0.9585168778896331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04148310944437981, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.9733283996582, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7667205810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1490, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.19971236884593963, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.11511102095246314, + "grad_norm": 0.1958649903535843, + "kept_token_count": 655.003125, + "learning_rate": 2.998e-05, + "loss": 0.0431, + "mgd_cos_sim": 0.9568580567836762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.043141933903098104, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.21426010131836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9133056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1500, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.20049205273389817, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.10927725359797477, + "grad_norm": 0.16023658215999603, + "kept_token_count": 653.85, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0434, + "mgd_cos_sim": 0.9566315770149231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.043368416279554366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.06335220336913, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9740844726563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1510, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.1994377076625824, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.10678720399737358, + "grad_norm": 0.12279165536165237, + "kept_token_count": 654.9890625, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0427, + "mgd_cos_sim": 0.9573456287384033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04265438318252564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 84.71053466796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8578399658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1520, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20065531879663467, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.10396389812231063, + "grad_norm": 0.17965103685855865, + "kept_token_count": 654.271875, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0395, + "mgd_cos_sim": 0.9605352222919464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03946477137506008, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 86.69228057861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8946105957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1530, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19961453676223756, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.11490071415901185, + "grad_norm": 0.19924460351467133, + "kept_token_count": 654.984375, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0425, + "mgd_cos_sim": 0.9574579477310181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.04254204370081425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 87.9146614074707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9884490966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1540, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19963298887014388, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.11331089064478875, + "grad_norm": 0.17979039251804352, + "kept_token_count": 655.2875, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0381, + "mgd_cos_sim": 0.961892181634903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03810781911015511, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 89.09564208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47142028808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1550, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.19987230002880096, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.11695291996002197, + "grad_norm": 0.13446369767189026, + "kept_token_count": 654.5578125, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0384, + "mgd_cos_sim": 0.9615664541721344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03843353800475598, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 90.32811279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.28049926757814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1560, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.19955074191093444, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.12155270799994469, + "grad_norm": 0.18085509538650513, + "kept_token_count": 655.0453125, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0396, + "mgd_cos_sim": 0.9604159772396088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.039584023877978325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.53780517578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.32109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1570, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20063437521457672, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.12601077407598496, + "grad_norm": 0.15571705996990204, + "kept_token_count": 653.9203125, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0398, + "mgd_cos_sim": 0.9602361142635345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03976387679576874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.33592071533204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9552703857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1580, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.19988367110490798, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.11601484715938568, + "grad_norm": 0.14702339470386505, + "kept_token_count": 654.8828125, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0384, + "mgd_cos_sim": 0.9616191327571869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03838086687028408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 91.40384445190429, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1247894287109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1590, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19977646172046662, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.1198851853609085, + "grad_norm": 0.1696116179227829, + "kept_token_count": 654.8734375, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0381, + "mgd_cos_sim": 0.9618893384933471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03811066225171089, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 92.54290008544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1991851806641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1600, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.1998392090201378, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.1101453572511673, + "grad_norm": 0.18971870839595795, + "kept_token_count": 654.7359375, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0385, + "mgd_cos_sim": 0.9614696264266968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.038530367985367776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 93.2064437866211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9254638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1610, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.1994919404387474, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.11356999203562737, + "grad_norm": 0.13069497048854828, + "kept_token_count": 654.740625, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0385, + "mgd_cos_sim": 0.9614950954914093, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.038504913076758386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.29778747558593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.69683837890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1620, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.1997174322605133, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.11876505613327026, + "grad_norm": 0.11138927191495895, + "kept_token_count": 654.675, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0378, + "mgd_cos_sim": 0.9622282683849335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.037771726958453655, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.033544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22218017578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.1999845564365387, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.11451392695307731, + "grad_norm": 0.20164676010608673, + "kept_token_count": 654.7015625, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0366, + "mgd_cos_sim": 0.9634267151355743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.036573294550180435, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.17841415405273, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2682830810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19888928532600403, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.11460494995117188, + "grad_norm": 0.22178886830806732, + "kept_token_count": 655.4265625, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0388, + "mgd_cos_sim": 0.9612253189086915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03877468481659889, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 96.26466217041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1273986816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.20011964291334153, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.11793006956577301, + "grad_norm": 0.22931866347789764, + "kept_token_count": 654.4921875, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0373, + "mgd_cos_sim": 0.9627173185348511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.037282684817910194, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.50500411987305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6542694091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1660, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19871231764554978, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.11734660640358925, + "grad_norm": 0.1789148449897766, + "kept_token_count": 655.9484375, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0399, + "mgd_cos_sim": 0.960106098651886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03989391513168812, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.6371452331543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55625915527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1670, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.200582292675972, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.10254736542701721, + "grad_norm": 0.23068441450595856, + "kept_token_count": 654.1984375, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0378, + "mgd_cos_sim": 0.9622122287750244, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.037787767872214315, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 95.47269744873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4118255615234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1680, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20000685304403304, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.11739195510745049, + "grad_norm": 0.17660480737686157, + "kept_token_count": 654.4953125, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0387, + "mgd_cos_sim": 0.9612572669982911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03874275535345077, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 94.61800079345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.44439086914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.1990959420800209, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.12085535004734993, + "grad_norm": 0.16673581302165985, + "kept_token_count": 655.3921875, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0377, + "mgd_cos_sim": 0.962301105260849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03769888263195753, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 97.55979232788086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.58738708496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2003359615802765, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.11069878190755844, + "grad_norm": 0.10532862693071365, + "kept_token_count": 654.315625, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0372, + "mgd_cos_sim": 0.9627739369869233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.037226077541708945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 98.1463996887207, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5564239501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1710, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20033180117607116, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.11559355482459069, + "grad_norm": 0.20529811084270477, + "kept_token_count": 654.21875, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0346, + "mgd_cos_sim": 0.9653689205646515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.034631097316741945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 99.72191696166992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.51576843261716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1720, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19962474554777146, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.11578413546085357, + "grad_norm": 0.12430015206336975, + "kept_token_count": 654.8390625, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0383, + "mgd_cos_sim": 0.961746096611023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.038253898918628695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 100.89799270629882, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.640771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1730, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19897158294916154, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.10509697124361991, + "grad_norm": 0.15444913506507874, + "kept_token_count": 655.4703125, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0374, + "mgd_cos_sim": 0.9626320004463196, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03736801818013191, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 101.21110916137695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55739440917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1740, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20027378350496292, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.1223619446158409, + "grad_norm": 0.1334921270608902, + "kept_token_count": 654.4046875, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0373, + "mgd_cos_sim": 0.9626731514930725, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03732684478163719, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.64293060302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70094604492186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1750, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2002595543861389, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.10986650586128235, + "grad_norm": 0.12536704540252686, + "kept_token_count": 654.5453125, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0341, + "mgd_cos_sim": 0.9659357607364655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.034064242243766786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 105.0649024963379, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9864044189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1760, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.19858355820178986, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.11796955540776252, + "grad_norm": 0.12776975333690643, + "kept_token_count": 655.5171875, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0376, + "mgd_cos_sim": 0.9623978137969971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03760218787938356, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.15287475585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1615753173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1770, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.19990862756967545, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.11711848899722099, + "grad_norm": 0.12827590107917786, + "kept_token_count": 654.7234375, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0344, + "mgd_cos_sim": 0.9655710518360138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03442893996834755, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.06340255737305, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.06533203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1780, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.1992767110466957, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.11395491361618042, + "grad_norm": 0.14395558834075928, + "kept_token_count": 655.1265625, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.034, + "mgd_cos_sim": 0.9660360097885132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0339639974758029, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 104.02651672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.33912048339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1790, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.20010918825864793, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.11804831475019455, + "grad_norm": 0.14701323211193085, + "kept_token_count": 654.53125, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0347, + "mgd_cos_sim": 0.9653411507606506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03465885240584612, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 105.79209365844727, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.89097900390624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1800, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2001691147685051, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.10738789364695549, + "grad_norm": 0.12912698090076447, + "kept_token_count": 654.4015625, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0341, + "mgd_cos_sim": 0.965861701965332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03413829617202282, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 103.0083999633789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1810, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.19978724420070648, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.11449096649885178, + "grad_norm": 0.12057028710842133, + "kept_token_count": 654.8078125, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0366, + "mgd_cos_sim": 0.9633510768413543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.036648920923471454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 103.2098617553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2903747558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1820, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2004532352089882, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.10542062893509865, + "grad_norm": 0.11513040214776993, + "kept_token_count": 654.5375, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0351, + "mgd_cos_sim": 0.96487375497818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03512624092400074, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 105.28628616333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25274658203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1830, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20118633061647415, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.11727861315011978, + "grad_norm": 0.22726809978485107, + "kept_token_count": 653.7234375, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0364, + "mgd_cos_sim": 0.9635651707649231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03643482699990273, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 107.58115005493164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57801513671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1840, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.1996232196688652, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.12504248991608619, + "grad_norm": 0.1610870361328125, + "kept_token_count": 654.7484375, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0366, + "mgd_cos_sim": 0.9633561909198761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.036643822118639945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 109.40990295410157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2359893798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1850, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20053764134645463, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.11397986188530922, + "grad_norm": 0.20490886270999908, + "kept_token_count": 654.0078125, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0365, + "mgd_cos_sim": 0.9634794056415558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03652059733867645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 107.48342971801758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2770080566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1860, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.19992830753326415, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.11305897608399391, + "grad_norm": 0.13990676403045654, + "kept_token_count": 654.7140625, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0332, + "mgd_cos_sim": 0.9667696595191956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.033230352029204366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 108.9787467956543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4966552734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1870, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20003028064966202, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.11953292861580848, + "grad_norm": 0.1374102234840393, + "kept_token_count": 654.71875, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0333, + "mgd_cos_sim": 0.966727477312088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03327252101153135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 107.92493133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8548980712891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1880, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20023452639579772, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.12710456550121307, + "grad_norm": 0.1373687982559204, + "kept_token_count": 654.6015625, + "learning_rate": 2.998621191519629e-05, + "loss": 0.035, + "mgd_cos_sim": 0.9649883270263672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03501168061047792, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 109.43601989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7437377929688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1890, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.19941204637289048, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.11842770725488663, + "grad_norm": 0.11917968839406967, + "kept_token_count": 655.171875, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0351, + "mgd_cos_sim": 0.9649186909198761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03508129678666592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.22200012207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4621520996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1900, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20041563808918, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.11533128395676613, + "grad_norm": 0.13053807616233826, + "kept_token_count": 654.29375, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0335, + "mgd_cos_sim": 0.9665318191051483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03346818946301937, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 111.88628921508788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.51466064453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.1995115488767624, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.11775240749120712, + "grad_norm": 0.13937196135520935, + "kept_token_count": 654.8734375, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0352, + "mgd_cos_sim": 0.9648016154766083, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.035198378190398213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 109.47786560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7388061523437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.1999636098742485, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.11716500520706177, + "grad_norm": 0.19092422723770142, + "kept_token_count": 654.371875, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0343, + "mgd_cos_sim": 0.9657262086868286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03427379354834557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 108.70901412963867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22764892578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1930, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.19999708235263824, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.11194603443145752, + "grad_norm": 0.12312964349985123, + "kept_token_count": 654.628125, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0331, + "mgd_cos_sim": 0.9668756484985351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0331243596971035, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 109.83605194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4172668457031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20048478096723557, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.11073949784040452, + "grad_norm": 0.11385220289230347, + "kept_token_count": 654.2125, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0314, + "mgd_cos_sim": 0.968616497516632, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03138350825756788, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 110.23999633789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6097473144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1950, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2004660502076149, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.11497483626008034, + "grad_norm": 0.1168496310710907, + "kept_token_count": 654.29375, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0295, + "mgd_cos_sim": 0.9704523086547852, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02954768780618906, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 111.1070571899414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0638031005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1960, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20127806365489959, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.1270100384950638, + "grad_norm": 0.15805818140506744, + "kept_token_count": 653.615625, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0321, + "mgd_cos_sim": 0.9679470896720886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03205291945487261, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 113.61663055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1523101806641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1970, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19973127394914628, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.12342466488480568, + "grad_norm": 0.12861698865890503, + "kept_token_count": 655.0203125, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0338, + "mgd_cos_sim": 0.9662430047988891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.033756998181343076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 111.24387969970704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.22822265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1980, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.1998128831386566, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.11833923757076263, + "grad_norm": 0.14025793969631195, + "kept_token_count": 654.8015625, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0342, + "mgd_cos_sim": 0.965845400094986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03415461536496878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 112.60590972900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0497985839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1990, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.19915823638439178, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.1110913373529911, + "grad_norm": 0.1341957300901413, + "kept_token_count": 655.3015625, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.031, + "mgd_cos_sim": 0.968954598903656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0310453986749053, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.2223503112793, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6488861083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2000, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19867531806230546, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.1275704562664032, + "grad_norm": 0.11647408455610275, + "kept_token_count": 655.5203125, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0342, + "mgd_cos_sim": 0.9658108055591583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03418919648975134, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 114.15258712768555, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0139556884766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.20010946244001387, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.11705309301614761, + "grad_norm": 0.10940506309270859, + "kept_token_count": 654.4578125, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0312, + "mgd_cos_sim": 0.9687553644180298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0312446478754282, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 114.21510848999023, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9301513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2020, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2006191208958626, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.1238844595849514, + "grad_norm": 0.103978730738163, + "kept_token_count": 653.9953125, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0319, + "mgd_cos_sim": 0.9681300640106201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03186994027346372, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.5526008605957, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81190185546876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2030, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19931214451789855, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.11385244801640511, + "grad_norm": 0.15181314945220947, + "kept_token_count": 655.0875, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0316, + "mgd_cos_sim": 0.9683501660823822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03164982814341784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.3658317565918, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.09388732910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2040, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20123942643404008, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.12482921779155731, + "grad_norm": 0.12129469215869904, + "kept_token_count": 653.578125, + "learning_rate": 2.99725410813577e-05, + "loss": 0.032, + "mgd_cos_sim": 0.967952823638916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.032047167606651786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.74608840942383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59495239257814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19991382360458373, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.1117615632712841, + "grad_norm": 0.1414732187986374, + "kept_token_count": 654.6828125, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0294, + "mgd_cos_sim": 0.9705772817134857, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02942272573709488, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.6728904724121, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47144470214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2060, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.1999826267361641, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.1357294276356697, + "grad_norm": 0.10617949813604355, + "kept_token_count": 654.4984375, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0347, + "mgd_cos_sim": 0.9653067529201508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03469325769692659, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.8801139831543, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7160278320313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2070, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20042162239551545, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.11105611398816109, + "grad_norm": 0.14905503392219543, + "kept_token_count": 654.31875, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0322, + "mgd_cos_sim": 0.9677950263023376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.032204960472881794, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 115.96807708740235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9616180419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2080, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19950230717658995, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.11906611025333405, + "grad_norm": 0.10697383433580399, + "kept_token_count": 654.8046875, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.031, + "mgd_cos_sim": 0.9690308392047882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03096915427595377, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.54153671264649, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.44920349121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2090, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.19908468425273895, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.12575173005461693, + "grad_norm": 0.1109326183795929, + "kept_token_count": 655.5171875, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0319, + "mgd_cos_sim": 0.9681313395500183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03186865486204624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 116.26914138793946, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9068603515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2100, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.199793304502964, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.12022633776068688, + "grad_norm": 0.1406773328781128, + "kept_token_count": 654.84375, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0326, + "mgd_cos_sim": 0.9674493193626403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0325506754219532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.47429885864258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5276733398438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2110, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20061843395233153, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.12133865058422089, + "grad_norm": 0.11333956569433212, + "kept_token_count": 654.1859375, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.029, + "mgd_cos_sim": 0.9709540665149688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029045926220715047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 120.00400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0658813476563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2120, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.1998453587293625, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.11673985645174981, + "grad_norm": 0.1401137262582779, + "kept_token_count": 654.8984375, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0339, + "mgd_cos_sim": 0.9661480247974396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03385197296738625, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 118.5062026977539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7842102050781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2130, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.19951820224523545, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.12042758539319039, + "grad_norm": 0.15172135829925537, + "kept_token_count": 654.859375, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0305, + "mgd_cos_sim": 0.9695453763008117, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03045461904257536, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 119.54155044555664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.07422790527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.20037884265184402, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.11539027541875839, + "grad_norm": 0.161556676030159, + "kept_token_count": 654.4515625, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0293, + "mgd_cos_sim": 0.9707021474838257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02929785531014204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 121.83653564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.88300170898435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2150, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.20053426623344422, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.11318441927433014, + "grad_norm": 0.1571590155363083, + "kept_token_count": 654.265625, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0296, + "mgd_cos_sim": 0.9704224169254303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02957757506519556, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.68419494628907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.15668029785155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2160, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20027921199798585, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.11166008040308953, + "grad_norm": 0.11142795532941818, + "kept_token_count": 654.2609375, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0286, + "mgd_cos_sim": 0.9714193403720855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02858067210763693, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.45322799682617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7128112792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2170, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20028292387723923, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.1209936112165451, + "grad_norm": 0.1599886268377304, + "kept_token_count": 654.4625, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0313, + "mgd_cos_sim": 0.9687092006206512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03129082098603249, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 120.80728302001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4650177001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2180, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20004068017005922, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.11076322421431542, + "grad_norm": 0.11550405621528625, + "kept_token_count": 654.415625, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0304, + "mgd_cos_sim": 0.9695701956748962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.030429807864129542, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 119.84375610351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0913970947266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.200064554810524, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.13263248950242995, + "grad_norm": 0.11277804523706436, + "kept_token_count": 654.7734375, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0301, + "mgd_cos_sim": 0.9698540389537811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.030145954340696335, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.00737838745117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83472595214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2200, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.20030218362808228, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.1207246132194996, + "grad_norm": 0.11909087747335434, + "kept_token_count": 654.6234375, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0292, + "mgd_cos_sim": 0.9707894444465637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029210562631487847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.90189361572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9350158691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2210, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.2006099060177803, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.12133844271302223, + "grad_norm": 0.11912630498409271, + "kept_token_count": 654.25625, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0282, + "mgd_cos_sim": 0.9717721521854401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.028227852284908296, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.22916259765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5102905273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2220, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.2002684772014618, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.12234652489423752, + "grad_norm": 0.18123695254325867, + "kept_token_count": 654.58125, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0312, + "mgd_cos_sim": 0.9688034772872924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.031196523644030093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.53384475708008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5271820068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2230, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.1991616502404213, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.12370899617671967, + "grad_norm": 0.12583531439304352, + "kept_token_count": 655.48125, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0297, + "mgd_cos_sim": 0.9702904939651489, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02970948740839958, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.9437355041504, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.84593200683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2240, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.19994312822818755, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.11871435195207596, + "grad_norm": 0.15389367938041687, + "kept_token_count": 654.6609375, + "learning_rate": 2.994890384673426e-05, + "loss": 0.031, + "mgd_cos_sim": 0.9689571917057037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.031042823754251003, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 122.34011993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.73094787597654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2000064432621002, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.12090981304645539, + "grad_norm": 0.10747815668582916, + "kept_token_count": 654.5828125, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0291, + "mgd_cos_sim": 0.9709112048149109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02908879444003105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 121.77363052368165, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.70537719726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2260, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19957653433084488, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.1103884033858776, + "grad_norm": 0.141089528799057, + "kept_token_count": 654.8140625, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0292, + "mgd_cos_sim": 0.9707635700702667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029236437380313875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.68569412231446, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5427947998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2270, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.2001085251569748, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.10979426354169845, + "grad_norm": 0.11270535737276077, + "kept_token_count": 654.5140625, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0295, + "mgd_cos_sim": 0.9704706013202667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029529397375881672, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.01393814086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7153259277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2280, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20019243955612182, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.12069524750113488, + "grad_norm": 0.09156475216150284, + "kept_token_count": 654.403125, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0293, + "mgd_cos_sim": 0.9707218945026398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02927810363471508, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.54983825683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7613525390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20018991976976394, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.12297803461551667, + "grad_norm": 0.12882131338119507, + "kept_token_count": 654.19375, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0302, + "mgd_cos_sim": 0.9697664022445679, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.03023359104990959, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 123.52852630615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9816101074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2300, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.20038278996944428, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.12398257553577423, + "grad_norm": 0.11875156313180923, + "kept_token_count": 654.553125, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0295, + "mgd_cos_sim": 0.9705447196960449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02945528570562601, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 124.0901596069336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6746887207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2310, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.1995249345898628, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.11718225404620171, + "grad_norm": 0.10345808416604996, + "kept_token_count": 654.990625, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0287, + "mgd_cos_sim": 0.9713359355926514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.028664061054587364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 125.81248550415039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1416412353516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19998779892921448, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.12149094343185425, + "grad_norm": 0.11534345895051956, + "kept_token_count": 654.5671875, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0294, + "mgd_cos_sim": 0.9706342399120331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029365762136876583, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.35680084228515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2571502685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2330, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19962584227323532, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.1217643640935421, + "grad_norm": 0.12005317956209183, + "kept_token_count": 655.1046875, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0297, + "mgd_cos_sim": 0.9703434765338897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029656535387039183, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 128.7586540222168, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.617578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19859777092933656, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.13174197003245353, + "grad_norm": 0.14428167045116425, + "kept_token_count": 655.58125, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0294, + "mgd_cos_sim": 0.9705838203430176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029416179470717906, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.1664192199707, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.27293090820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2350, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.20009984374046325, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.12210036292672158, + "grad_norm": 0.1413341760635376, + "kept_token_count": 654.5125, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0292, + "mgd_cos_sim": 0.9708192765712738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029180728644132615, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 125.83987731933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.775048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.19928498268127443, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.12100033909082412, + "grad_norm": 0.09406057000160217, + "kept_token_count": 655.1875, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0298, + "mgd_cos_sim": 0.9702435493469238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.029756446927785875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 127.68882751464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7957397460938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2370, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.1987648442387581, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.11187446638941764, + "grad_norm": 0.07778773456811905, + "kept_token_count": 655.7296875, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0267, + "mgd_cos_sim": 0.9733031988143921, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026696803420782088, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.01985244750976, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6886047363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2380, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2002652496099472, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.12568045556545257, + "grad_norm": 0.13124091923236847, + "kept_token_count": 654.740625, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0264, + "mgd_cos_sim": 0.9735882699489593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02641171757131815, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 125.70115966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41636047363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2390, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.20097984373569489, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.11775615811347961, + "grad_norm": 0.08996540307998657, + "kept_token_count": 654.0546875, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0271, + "mgd_cos_sim": 0.9728940784931183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02710591871291399, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 127.78148574829102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07142639160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2400, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.20112920850515364, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.11510329768061638, + "grad_norm": 0.11855271458625793, + "kept_token_count": 653.728125, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9738242745399475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026175707764923572, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 130.85702667236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.53008728027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2410, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.20015017688274384, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.1257094644010067, + "grad_norm": 0.11453552544116974, + "kept_token_count": 654.328125, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0285, + "mgd_cos_sim": 0.9715283036231994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.028471715562045575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 129.70748748779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7334686279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2420, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.20036747455596923, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.11975226998329162, + "grad_norm": 0.14426769316196442, + "kept_token_count": 654.103125, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0277, + "mgd_cos_sim": 0.9722967028617859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027703293412923814, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 130.34992218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5820587158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2430, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.19932838827371596, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.11310407370328904, + "grad_norm": 0.11939860880374908, + "kept_token_count": 655.3265625, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0272, + "mgd_cos_sim": 0.9728412985801697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027158707752823828, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.88032836914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9342010498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2440, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.1999538794159889, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.13271308913826943, + "grad_norm": 0.1587144434452057, + "kept_token_count": 654.8890625, + "learning_rate": 2.991800117455e-05, + "loss": 0.0276, + "mgd_cos_sim": 0.97235067486763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02764932494610548, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 129.57857437133788, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5327911376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2450, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.19991170167922973, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.12072004526853561, + "grad_norm": 0.10284410417079926, + "kept_token_count": 655.0921875, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0275, + "mgd_cos_sim": 0.972450464963913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027549535594880582, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 126.68024368286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2113311767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2460, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.1995989352464676, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.13087283447384834, + "grad_norm": 0.15565797686576843, + "kept_token_count": 655.3625, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0279, + "mgd_cos_sim": 0.9720678389072418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027932167612016202, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 131.06892166137695, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5684326171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2470, + "valid_token_count": 818.7953125 + }, + { + "actual_token_mask_ratio": 0.19974400997161865, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.11484827473759651, + "grad_norm": 0.12248361110687256, + "kept_token_count": 654.8515625, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0272, + "mgd_cos_sim": 0.9728238880634308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027176103368401528, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 133.14500427246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.52063903808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2480, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.20039330720901488, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.1147394098341465, + "grad_norm": 0.11864753812551498, + "kept_token_count": 654.4484375, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9741387724876404, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025861238501965998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.9502182006836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0565521240234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2490, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.20067113488912583, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.12447682097554207, + "grad_norm": 0.12532776594161987, + "kept_token_count": 654.0859375, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0286, + "mgd_cos_sim": 0.9713739037513733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.028626099787652494, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 134.77554626464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.06651000976564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2500, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20062699317932128, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.12460567876696586, + "grad_norm": 0.12449522316455841, + "kept_token_count": 654.2984375, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0286, + "mgd_cos_sim": 0.9713893532752991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02861065175384283, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 134.19266662597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30772399902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2510, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19969678670167923, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.1233951523900032, + "grad_norm": 0.11282117664813995, + "kept_token_count": 654.784375, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0273, + "mgd_cos_sim": 0.9726986229419708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027301381342113018, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 133.1185562133789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8155914306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1996464267373085, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.12472715079784394, + "grad_norm": 0.12680891156196594, + "kept_token_count": 654.99375, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.973768663406372, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026231345906853675, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 130.23902969360353, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2268829345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2530, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.20038108080625533, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.11321731507778168, + "grad_norm": 0.12521451711654663, + "kept_token_count": 654.1578125, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0269, + "mgd_cos_sim": 0.9731282234191895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02687178961932659, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.01536560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.39117736816405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2540, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19931571632623674, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.12919616773724557, + "grad_norm": 0.13533927500247955, + "kept_token_count": 654.9140625, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0278, + "mgd_cos_sim": 0.9721893906593323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027810598351061346, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 131.72611389160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.88103637695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2550, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.19928652048110962, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.1214998297393322, + "grad_norm": 0.11387769877910614, + "kept_token_count": 655.2140625, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0284, + "mgd_cos_sim": 0.9715895652770996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.028410438261926175, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 134.2849136352539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4589111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2560, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20094604790210724, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.1157404825091362, + "grad_norm": 0.11860650777816772, + "kept_token_count": 653.996875, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9738218426704407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026178159192204474, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 136.59978485107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4284973144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2570, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.1991339161992073, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.11735129058361053, + "grad_norm": 0.14311377704143524, + "kept_token_count": 655.5453125, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9740785300731659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02592147532850504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 135.17953796386718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.50543212890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2580, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.2001371204853058, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.10975858867168427, + "grad_norm": 0.11967255175113678, + "kept_token_count": 654.434375, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0266, + "mgd_cos_sim": 0.9733564257621765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026643577218055724, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 135.09981536865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3965209960937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2590, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19960513412952424, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.12579987868666648, + "grad_norm": 0.08948090672492981, + "kept_token_count": 654.978125, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0267, + "mgd_cos_sim": 0.9733450174331665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026654988154768945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.54790267944335, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1888397216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2600, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.1996015191078186, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.11128647401928901, + "grad_norm": 0.11598998308181763, + "kept_token_count": 654.8890625, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9743611216545105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025638875737786292, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 133.55074768066407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.587841796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2610, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.20009875893592835, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.1074554018676281, + "grad_norm": 0.12452611327171326, + "kept_token_count": 654.2984375, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9738519370555878, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026148055866360664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 132.033837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19226989746096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2620, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20031544119119643, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.11412117406725883, + "grad_norm": 0.09279025346040726, + "kept_token_count": 654.3046875, + "learning_rate": 2.988398916103648e-05, + "loss": 0.025, + "mgd_cos_sim": 0.974975723028183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025024271011352538, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 135.02970428466796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5246612548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2630, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.20063410997390746, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.11914138197898864, + "grad_norm": 0.12197666615247726, + "kept_token_count": 654.3703125, + "learning_rate": 2.988192766402212e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9739950120449066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0260049756616354, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.5491912841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47522888183596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2640, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.19974142611026763, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.12436557039618493, + "grad_norm": 0.13978907465934753, + "kept_token_count": 654.734375, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9744712173938751, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02552878614515066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.95218505859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7669372558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20020836293697358, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.12480978295207024, + "grad_norm": 0.1727427989244461, + "kept_token_count": 654.359375, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0271, + "mgd_cos_sim": 0.9728940427303314, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.027105961926281452, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 136.8648651123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15468139648436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2660, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.19967750310897828, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.11882718652486801, + "grad_norm": 0.11768099665641785, + "kept_token_count": 654.6828125, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0279, + "mgd_cos_sim": 0.972063410282135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02793658636510372, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 138.48170013427733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1087677001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2670, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20025371462106706, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.12120778039097786, + "grad_norm": 0.1290924996137619, + "kept_token_count": 654.3734375, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9750450849533081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02495491448789835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 135.04093322753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48369750976565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2680, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2003630816936493, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.11916580498218536, + "grad_norm": 0.10270458459854126, + "kept_token_count": 654.5, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9750366032123565, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.024963388033211232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.0094985961914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.24900512695314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2690, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.19951732605695724, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.1323087178170681, + "grad_norm": 0.1260640025138855, + "kept_token_count": 654.784375, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9736610054969788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026339004375040533, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.39694061279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0809844970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2700, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.20063746869564056, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.12882199808955191, + "grad_norm": 0.12835098803043365, + "kept_token_count": 654.190625, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9746540009975433, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025346001610159874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.3652770996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0251220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2710, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1991313487291336, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.12646219432353972, + "grad_norm": 0.09703364968299866, + "kept_token_count": 655.1796875, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0267, + "mgd_cos_sim": 0.9733107388019562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02668926361948252, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 141.24376373291017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25708923339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2002478525042534, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.1220293365418911, + "grad_norm": 0.16507278382778168, + "kept_token_count": 654.39375, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9738984763622284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026101524569094182, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 136.17954864501954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.13877258300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2730, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1990777775645256, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.12015454769134522, + "grad_norm": 0.10735068470239639, + "kept_token_count": 655.128125, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9751949310302734, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.024805046804249286, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.78815155029298, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.79552001953124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2740, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.19932498186826705, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.11578205302357673, + "grad_norm": 0.09021393209695816, + "kept_token_count": 655.0125, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9750062763690949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02499369941651821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.4389434814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2405090332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2750, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.19999905526638032, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.1172834500670433, + "grad_norm": 0.09509250521659851, + "kept_token_count": 654.709375, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9738389134407044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026161086931824683, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.80525665283204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9221252441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2760, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.1993798077106476, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.12101110517978668, + "grad_norm": 0.10859014838933945, + "kept_token_count": 654.828125, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9758637130260468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.024136289954185486, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 139.95425567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.96640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2770, + "valid_token_count": 817.9015625 + }, + { + "actual_token_mask_ratio": 0.20007502883672715, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.12155347391963005, + "grad_norm": 0.11795797944068909, + "kept_token_count": 654.1953125, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9749355316162109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025064455717802046, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 140.5051528930664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4063232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2780, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.1991599276661873, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.11903155818581582, + "grad_norm": 0.10322503745555878, + "kept_token_count": 655.1828125, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9738900482654571, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026109957508742808, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.91200561523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3613006591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19983706027269363, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.10802164450287818, + "grad_norm": 0.10528305917978287, + "kept_token_count": 654.496875, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9760903120040894, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023909684084355832, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.91802825927735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2536224365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2800, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.2002142995595932, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.13529117703437804, + "grad_norm": 0.14245523512363434, + "kept_token_count": 654.6015625, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9736727774143219, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026327226869761945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 141.65985107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2810, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.2000141978263855, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.1242129571735859, + "grad_norm": 0.1292983591556549, + "kept_token_count": 654.6734375, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9740667879581452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0259332150220871, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 137.96505889892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.55934448242186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2820, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19856786876916885, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.12193970680236817, + "grad_norm": 0.1386420875787735, + "kept_token_count": 655.9859375, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9764565646648407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02354343477636576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 142.98818817138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.38269958496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2830, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.19935524165630342, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.11760799139738083, + "grad_norm": 0.12722796201705933, + "kept_token_count": 655.175, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.975289660692215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0247103426605463, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.93717193603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3918792724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2840, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.20137961506843566, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.12086788043379784, + "grad_norm": 0.0879160687327385, + "kept_token_count": 653.5375, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9766086280345917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02339137401431799, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.66521148681642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.35882873535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2850, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20005265921354293, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.12786465659737586, + "grad_norm": 0.11390845477581024, + "kept_token_count": 654.7640625, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9766570508480072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02334296442568302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.26519622802735, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.17853088378905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2860, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20071848034858703, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.12088745534420013, + "grad_norm": 0.10105278342962265, + "kept_token_count": 653.8609375, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9767193734645844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023280623927712442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.68350830078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57567138671874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2870, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.1998555392026901, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.11797314658761024, + "grad_norm": 0.08645257353782654, + "kept_token_count": 654.615625, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9743893444538116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025610643438994884, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.33375701904296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94398498535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2880, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.1985640436410904, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.12900796830654143, + "grad_norm": 0.1370023787021637, + "kept_token_count": 655.7875, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9735029458999633, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.026497058756649493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.0260437011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4858642578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2890, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19942215979099273, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.11544566452503205, + "grad_norm": 0.15458430349826813, + "kept_token_count": 654.871875, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9757296085357666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.024270392023026943, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.4418960571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4171569824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2900, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.19985919296741486, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.11924180686473847, + "grad_norm": 0.15673187375068665, + "kept_token_count": 654.921875, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9750012457370758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02499876096844673, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.50520935058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6365203857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2910, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.19982483386993408, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.12425901666283608, + "grad_norm": 0.11065347492694855, + "kept_token_count": 654.615625, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9738379061222077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02616208642721176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.9399658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9212890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.199113430082798, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.11718654111027718, + "grad_norm": 0.13813553750514984, + "kept_token_count": 655.6265625, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9747510492801666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025248966924846173, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.0759307861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93333435058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2930, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.20082511156797409, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.1254302941262722, + "grad_norm": 0.14307865500450134, + "kept_token_count": 654.0046875, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9761318147182465, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023868185095489027, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.55491943359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89284057617186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2940, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19925573021173476, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.11946597397327423, + "grad_norm": 0.11588966846466064, + "kept_token_count": 655.465625, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9761988580226898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023801135085523128, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.37247009277343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7471923828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2950, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.19948100596666335, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.12251722440123558, + "grad_norm": 0.08326196670532227, + "kept_token_count": 655.1859375, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9769866108894348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02301339078694582, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 146.87122802734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.64048461914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2960, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.19963623434305192, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.12394338473677635, + "grad_norm": 0.11690182983875275, + "kept_token_count": 654.91875, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.976296204328537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02370379399508238, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.13815002441407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3564910888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19992702603340148, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.13030113652348518, + "grad_norm": 0.13454659283161163, + "kept_token_count": 654.79375, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9748875021934509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.025112506560981275, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.57800903320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3017211914063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20030145794153215, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.11191362738609315, + "grad_norm": 0.09308222681283951, + "kept_token_count": 654.153125, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9760888397693634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023911166191101074, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.8053756713867, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94659729003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2990, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.20008822083473204, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.10940730422735215, + "grad_norm": 0.0991126298904419, + "kept_token_count": 654.88125, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9773351311683655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022664880380034447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.69371337890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0591705322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3000, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.20031243413686753, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.12022112384438514, + "grad_norm": 0.11561987549066544, + "kept_token_count": 654.4546875, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9764720499515533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023527939431369306, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.6525619506836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1178833007813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3010, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19866929501295089, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.12110088244080544, + "grad_norm": 0.1230693906545639, + "kept_token_count": 656.0890625, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.025, + "mgd_cos_sim": 0.974992436170578, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02500756215304136, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 145.7532699584961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.09955444335935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3020, + "valid_token_count": 818.7578125 + }, + { + "actual_token_mask_ratio": 0.2004692479968071, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.12048231661319733, + "grad_norm": 0.08077426999807358, + "kept_token_count": 654.4171875, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9768045485019684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023195459879934786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.38458251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.56541442871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3030, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.20000681579113005, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.1265024408698082, + "grad_norm": 0.0959751084446907, + "kept_token_count": 654.7765625, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9759619176387787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02403805833309889, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.21034545898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7502807617187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3040, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20014788210391998, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.11623304411768913, + "grad_norm": 0.11588701605796814, + "kept_token_count": 654.234375, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9766496658325196, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023350313864648344, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.0364776611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2996765136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3050, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.20047797113656998, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.11968627274036407, + "grad_norm": 0.12134752422571182, + "kept_token_count": 653.9375, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9749121963977814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02508780900388956, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.50720977783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.34234008789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3060, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.2008245900273323, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.11461328268051148, + "grad_norm": 0.11577158421278, + "kept_token_count": 653.9453125, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9779159367084503, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02208407036960125, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.26747436523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9063262939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3070, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.199129381775856, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.11773024201393127, + "grad_norm": 0.094362273812294, + "kept_token_count": 655.103125, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9766888320446014, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023311170935630798, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.57322692871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5530303955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20032330304384233, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.1204446330666542, + "grad_norm": 0.12158581614494324, + "kept_token_count": 654.2, + "learning_rate": 2.977048638861072e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9780371487140656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021962850354611874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.65398101806642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0603515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3090, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.1995035171508789, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.13298745974898338, + "grad_norm": 0.11163415759801865, + "kept_token_count": 654.8203125, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9763850450515748, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023614950850605963, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.55618743896486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9991180419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3100, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19953107386827468, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.1273059405386448, + "grad_norm": 0.10998232662677765, + "kept_token_count": 655.0703125, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9767542779445648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02324571870267391, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.4626495361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5261627197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3110, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20030439049005508, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.12033099681138992, + "grad_norm": 0.09601046144962311, + "kept_token_count": 654.4890625, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9771967649459838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022803233563899995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.31302490234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0287811279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3120, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.19968124181032182, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.12436500638723373, + "grad_norm": 0.12009298801422119, + "kept_token_count": 654.771875, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9774909853935242, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02250902131199837, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.54615173339843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.92930603027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3130, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19959430396556854, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.12243633046746254, + "grad_norm": 0.11418500542640686, + "kept_token_count": 654.946875, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9774968028068542, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022503208182752133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.8572006225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4530944824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3140, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19998055696487427, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.11762765422463417, + "grad_norm": 0.13521325588226318, + "kept_token_count": 654.7203125, + "learning_rate": 2.975287511396039e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9769860029220581, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02301399428397417, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.55501861572264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.15947265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3150, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.19909061938524247, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.12112505510449409, + "grad_norm": 0.12745775282382965, + "kept_token_count": 655.4828125, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9768700957298279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023129909858107568, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.02151489257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7613037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3160, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19922261387109758, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.1164622388780117, + "grad_norm": 0.09617628902196884, + "kept_token_count": 655.2546875, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9785922110080719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021407797746360303, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.17867736816407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.56703796386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3170, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19951021820306777, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.11917291656136512, + "grad_norm": 0.10907392203807831, + "kept_token_count": 655.1203125, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.977226710319519, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022773295268416406, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.8582290649414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2272033691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3180, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19941545724868776, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.12149692550301552, + "grad_norm": 0.12910735607147217, + "kept_token_count": 655.025, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9761475324630737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023852462135255336, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.72830352783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.14600524902346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3190, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19946496486663817, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.11786362156271935, + "grad_norm": 0.133872389793396, + "kept_token_count": 655.165625, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9765014708042145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023498537391424178, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.0291763305664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.09783935546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3200, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.20023611634969712, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.12324573174118995, + "grad_norm": 0.13151538372039795, + "kept_token_count": 654.403125, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9764614045619965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02353859469294548, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.61519317626954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53418884277346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1996507614850998, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.11725422218441964, + "grad_norm": 0.11101526767015457, + "kept_token_count": 654.825, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9778363049030304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022163693793118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.57898712158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5548522949219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3220, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20021891593933105, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.11340627670288086, + "grad_norm": 0.13013705611228943, + "kept_token_count": 654.4015625, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9791170418262481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020882967114448547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.09586334228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2787658691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3230, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2002668172121048, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.11577264070510865, + "grad_norm": 0.13190104067325592, + "kept_token_count": 654.109375, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9791574954986573, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02084251455962658, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.22007751464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.70143432617186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3240, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.1996366411447525, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.12243976220488548, + "grad_norm": 0.10631930828094482, + "kept_token_count": 654.99375, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9760822236537934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023917794972658158, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.81529998779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4764068603516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3250, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19983464181423188, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.12539364248514176, + "grad_norm": 0.09382365643978119, + "kept_token_count": 654.7109375, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9772853672504425, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022714646346867085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.15968475341796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.04717407226565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3260, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.20039032101631166, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.11791238635778427, + "grad_norm": 0.13871872425079346, + "kept_token_count": 654.5359375, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9781610667705536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021838935278356075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.81453857421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6940582275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3270, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1999578595161438, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.12992506772279738, + "grad_norm": 0.11587011069059372, + "kept_token_count": 654.56875, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9763806521892547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.023619342409074307, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.15672454833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49322814941405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3280, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20072492063045502, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.1329077146947384, + "grad_norm": 0.10010812431573868, + "kept_token_count": 654.0546875, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9768450379371643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02315498609095812, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.2724365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.279638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3290, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20015930384397507, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.12324942424893379, + "grad_norm": 0.08579578995704651, + "kept_token_count": 654.6015625, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9783757627010345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021624241769313813, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.55283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.55060424804685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3300, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.200461046397686, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.12178433164954186, + "grad_norm": 0.0942290723323822, + "kept_token_count": 654.2046875, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9792392730712891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020760720781981944, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.33447875976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.20819396972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3310, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19955755770206451, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.12564096972346306, + "grad_norm": 0.09599136561155319, + "kept_token_count": 655.053125, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9780861914157868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02191380150616169, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.06421508789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0706726074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3320, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20016487538814545, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.12345592752099037, + "grad_norm": 0.07742757350206375, + "kept_token_count": 654.428125, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.977804982662201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022195001877844334, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.80855407714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1627990722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3330, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20001964122056962, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.11656732559204101, + "grad_norm": 0.10247417539358139, + "kept_token_count": 654.3515625, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9785807132720947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021419283375144003, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.95169982910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4055572509766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3340, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.19960485696792601, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.12868774831295013, + "grad_norm": 0.08746465295553207, + "kept_token_count": 654.759375, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9780406057834625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02195940036326647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.56683807373048, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3844848632813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3350, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.19957961589097978, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.12012544646859169, + "grad_norm": 0.10809481143951416, + "kept_token_count": 654.7765625, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9779268145561218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02207318190485239, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.6704299926758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1868621826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3360, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.19956735670566558, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.12067336514592171, + "grad_norm": 0.12489312887191772, + "kept_token_count": 654.93125, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9773737132549286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022626291774213315, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.7401153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4241027832031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3370, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.20026648938655853, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.11753748878836631, + "grad_norm": 0.09841695427894592, + "kept_token_count": 654.4953125, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9785277485847473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02147223837673664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.55074157714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.41307373046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3380, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20041550248861312, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.1072719745337963, + "grad_norm": 0.0936804860830307, + "kept_token_count": 654.165625, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9778998494148254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02210014723241329, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.87601318359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.48850402832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3390, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.19875487685203552, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.12014180347323418, + "grad_norm": 0.10069993883371353, + "kept_token_count": 655.35, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9782120287418365, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021787958592176436, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.168603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66176147460936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3400, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.20048083513975143, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.10699792355298995, + "grad_norm": 0.15258991718292236, + "kept_token_count": 654.1078125, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9787447094917298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02125528361648321, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.6564743041992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2212158203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3410, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.1992398276925087, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.10432246178388596, + "grad_norm": 0.07847759872674942, + "kept_token_count": 655.21875, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9796960651874542, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02030393909662962, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 158.4006774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1317840576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3420, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2000698432326317, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.12443957775831223, + "grad_norm": 0.11281457543373108, + "kept_token_count": 654.4421875, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9784765720367432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02152341976761818, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.22349700927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6372467041016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3430, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.20087326020002366, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.12937405556440354, + "grad_norm": 0.09121129661798477, + "kept_token_count": 653.6203125, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9776135563850403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022386435978114606, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.41869049072267, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8126983642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3440, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.1996989518404007, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.12796833738684654, + "grad_norm": 0.08586462587118149, + "kept_token_count": 655.13125, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9787009060382843, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021299082785844803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.69102935791017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1374542236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3450, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.1995859682559967, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.126195178180933, + "grad_norm": 0.10369395464658737, + "kept_token_count": 654.6359375, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.979906040430069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02009396757930517, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.48773956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49462890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3460, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.20106572657823563, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.12335101813077927, + "grad_norm": 0.11987613141536713, + "kept_token_count": 653.8875, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9783815443515778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021618459187448025, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.36651306152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7420928955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3470, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19983987510204315, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.13186590299010276, + "grad_norm": 0.18281851708889008, + "kept_token_count": 655.05625, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9794878125190735, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020512193068861963, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.2837112426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4393249511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3480, + "valid_token_count": 818.6578125 + }, + { + "actual_token_mask_ratio": 0.20002565830945968, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.1154213584959507, + "grad_norm": 0.11140725761651993, + "kept_token_count": 654.6140625, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9790916204452514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020908390171825886, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.52061920166017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49449462890624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3490, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20014006942510604, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.1164268545806408, + "grad_norm": 0.0867757499217987, + "kept_token_count": 654.575, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9793745636940002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02062543872743845, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.46878662109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4374572753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3500, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.20057498812675476, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.12609345242381095, + "grad_norm": 0.10446397215127945, + "kept_token_count": 654.084375, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9787228167057037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02127719223499298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.76986999511718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.32085571289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3510, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19941187053918838, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.11353531777858734, + "grad_norm": 0.09504091739654541, + "kept_token_count": 655.1546875, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9796041011810303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02039589062333107, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 161.072119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0820343017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3520, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19931468963623047, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.12393944934010506, + "grad_norm": 0.14291591942310333, + "kept_token_count": 655.2734375, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9791051030158997, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020894877426326276, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.73427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83066711425784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.1998918443918228, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.11727245151996613, + "grad_norm": 0.09133854508399963, + "kept_token_count": 655.0734375, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9790646314620972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020935362204909325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.95823822021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.0952545166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3540, + "valid_token_count": 818.73125 + }, + { + "actual_token_mask_ratio": 0.20069065392017366, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.12552098855376242, + "grad_norm": 0.11218466609716415, + "kept_token_count": 654.0328125, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9787951946258545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021204806677997113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.69068298339843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1720458984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3550, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20050894021987914, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.11937059238553047, + "grad_norm": 0.08923203498125076, + "kept_token_count": 654.109375, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9792973279953003, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020702673867344857, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.37745666503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.96693420410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3560, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.1999846175312996, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.1143948070704937, + "grad_norm": 0.08454374969005585, + "kept_token_count": 654.5703125, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9794811904430389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02051881067454815, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 166.25828399658204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.95433349609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20090128779411315, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.12966737672686576, + "grad_norm": 0.08489885181188583, + "kept_token_count": 654.084375, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9779278755187988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.022072129510343076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.33074798583985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5017517089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3580, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.20053889751434326, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.13424732014536858, + "grad_norm": 0.08252883702516556, + "kept_token_count": 654.265625, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.979375684261322, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02062432188540697, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.82079467773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.31971740722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3590, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2002183973789215, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.11986973434686661, + "grad_norm": 0.09911278635263443, + "kept_token_count": 654.4703125, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9798443675041199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020155633240938185, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.43463439941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8728820800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3600, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.1992766723036766, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.12347394227981567, + "grad_norm": 0.0971018522977829, + "kept_token_count": 655.0046875, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9795365869998932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020463428646326064, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.2198699951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7316925048828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3610, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20075408667325972, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.12130676582455635, + "grad_norm": 0.09551296383142471, + "kept_token_count": 653.6515625, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9803942024707795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019605802744626998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.74861602783204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8054656982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3620, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.1993391752243042, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.12300967499613762, + "grad_norm": 0.08463121950626373, + "kept_token_count": 655.0046875, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9800273239612579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019972676411271095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.06100311279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41872253417966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3630, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19970421642065048, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.11461201831698417, + "grad_norm": 0.09492802619934082, + "kept_token_count": 655.0984375, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9795800745487213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020419909432530404, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.73270568847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6978302001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3640, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.19909008890390395, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.12858448177576065, + "grad_norm": 0.08944227546453476, + "kept_token_count": 655.290625, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9791000962257386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020899910666048526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.95138244628907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8930725097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3650, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.20071164816617965, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.12458944320678711, + "grad_norm": 0.08599881827831268, + "kept_token_count": 654.040625, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9796446681022644, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02035532183945179, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.07717742919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5966094970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20049312114715576, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.12259815633296967, + "grad_norm": 0.08763186633586884, + "kept_token_count": 654.234375, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9789599895477294, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021040017530322076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.53381958007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9344940185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3670, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20000472962856292, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.11883242279291154, + "grad_norm": 0.12461883574724197, + "kept_token_count": 654.903125, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9781557440757751, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.021844263188540936, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 171.5565612792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3329620361328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3680, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.19960099309682847, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.12882929369807244, + "grad_norm": 0.11689503490924835, + "kept_token_count": 654.9328125, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9791783452033996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020821655727922918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.47283325195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1963348388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3690, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19999131709337234, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.12138326466083527, + "grad_norm": 0.10886776447296143, + "kept_token_count": 654.8953125, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9804760336875915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019523966405540705, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 171.2392608642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8410369873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3700, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.1997459352016449, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.12994734197854996, + "grad_norm": 0.08558917045593262, + "kept_token_count": 654.75, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9794698238372803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02053018193691969, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.44662170410157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.45652465820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3710, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2004832148551941, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.1262687101960182, + "grad_norm": 0.09409631788730621, + "kept_token_count": 654.325, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9792131304740905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020786877162754534, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 168.54415130615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8546539306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3720, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20002485662698746, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.11491968035697937, + "grad_norm": 0.10900787264108658, + "kept_token_count": 654.5875, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9794561684131622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020543847233057022, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.3321762084961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7166259765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3730, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20022715926170348, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.12442310899496078, + "grad_norm": 0.0914333164691925, + "kept_token_count": 654.4765625, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9792368352413178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020763183385133742, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.12178497314454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1487152099609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3740, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20006968677043915, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.1342130072414875, + "grad_norm": 0.08814749866724014, + "kept_token_count": 654.2140625, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9804386556148529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019561355002224445, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.65919647216796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.77582397460935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3750, + "valid_token_count": 817.840625 + }, + { + "actual_token_mask_ratio": 0.20000442415475844, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.1333664610981941, + "grad_norm": 0.08674661070108414, + "kept_token_count": 654.6046875, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9797663569450379, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02023365180939436, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.5792984008789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0050903320313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3760, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20006887316703797, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.12417795732617379, + "grad_norm": 0.09245999902486801, + "kept_token_count": 654.3640625, + "learning_rate": 2.953325994871975e-05, + "loss": 0.019, + "mgd_cos_sim": 0.981015932559967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018984062224626543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.2534622192383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7410949707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3770, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19952186197042465, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.12888666093349457, + "grad_norm": 0.0803861990571022, + "kept_token_count": 655.1203125, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9799322366714478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0200677621178329, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 171.0846160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01572875976564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3780, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.199742291867733, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.1232532098889351, + "grad_norm": 0.09963696449995041, + "kept_token_count": 654.7140625, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.981691551208496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01830843072384596, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.1369598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1560363769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3790, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2000648558139801, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.1314414121210575, + "grad_norm": 0.08405514061450958, + "kept_token_count": 654.4859375, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9801805317401886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019819462858140468, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.92625579833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35792541503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3800, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.1998010829091072, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.1258398100733757, + "grad_norm": 0.09757097065448761, + "kept_token_count": 654.575, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9795549213886261, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020445075817406178, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.6209228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7182586669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3810, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.20019346475601196, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.12400588318705559, + "grad_norm": 0.1393885612487793, + "kept_token_count": 654.453125, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9807649254798889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019235078990459443, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 174.5376449584961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.73182373046876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3820, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19959813207387925, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.11893481388688087, + "grad_norm": 0.10950153321027756, + "kept_token_count": 655.0859375, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9806176841259002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019382316246628763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.26839904785157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98973083496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3830, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.1994056448340416, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.12180605232715606, + "grad_norm": 0.08506423979997635, + "kept_token_count": 655.0640625, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9801005601882935, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019899437110871075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 174.73438415527343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6188110351562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3840, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.1988716959953308, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.12544775903224945, + "grad_norm": 0.12779918313026428, + "kept_token_count": 655.6015625, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9797914206981659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020208588056266307, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.3298583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.13849487304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3850, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.1992288798093796, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.12265618368983269, + "grad_norm": 0.09098570793867111, + "kept_token_count": 655.028125, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.979162871837616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020837112702429293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 174.5349853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4217498779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3860, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20062262564897537, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.12416069954633713, + "grad_norm": 0.0899161845445633, + "kept_token_count": 654.1, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9806961238384246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0193038834258914, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.12332000732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1962585449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3870, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2001537099480629, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.12076823562383651, + "grad_norm": 0.0892048180103302, + "kept_token_count": 654.4296875, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9808270335197449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019172976724803448, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.8011505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.616943359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3880, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.2002149134874344, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.13179853111505507, + "grad_norm": 0.10071798413991928, + "kept_token_count": 654.4421875, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9795193433761596, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02048065550625324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.03340911865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5488677978516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3890, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.1995885267853737, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.1260899618268013, + "grad_norm": 0.09595692902803421, + "kept_token_count": 655.096875, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9793164551258087, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02068354319781065, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.75060424804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.34214172363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3900, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20079571157693862, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.13201253935694696, + "grad_norm": 0.144252747297287, + "kept_token_count": 654.0640625, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9800012528896331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019998739287257195, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.13359527587892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1327667236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3910, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20054430067539214, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.13715484142303466, + "grad_norm": 0.09086170792579651, + "kept_token_count": 654.165625, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.980222201347351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019777807220816613, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.6141616821289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3887542724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3920, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19978946447372437, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.12475199401378631, + "grad_norm": 0.12315365672111511, + "kept_token_count": 654.3734375, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9812661588191987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01873383391648531, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.644970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.54005126953126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3930, + "valid_token_count": 817.759375 + }, + { + "actual_token_mask_ratio": 0.20123647153377533, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.1274283155798912, + "grad_norm": 0.10580483824014664, + "kept_token_count": 653.509375, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9800507605075837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019949247129261494, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.68030395507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4313690185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20036386251449584, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.127376276999712, + "grad_norm": 0.09398426860570908, + "kept_token_count": 654.26875, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9806533992290497, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019346598908305167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.76647186279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2945495605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3950, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19953108131885527, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.12343243509531021, + "grad_norm": 0.1092834621667862, + "kept_token_count": 654.903125, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9794509649276734, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.02054902371019125, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.31200714111327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7129699707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3960, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19995319247245788, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.12621259614825248, + "grad_norm": 0.08550291508436203, + "kept_token_count": 654.659375, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9802472949028015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01975270789116621, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 174.78375396728515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3892333984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19993160665035248, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.13412071093916894, + "grad_norm": 0.07942477613687515, + "kept_token_count": 654.7796875, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9804740428924561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019525942951440813, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.52488861083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.91798706054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3980, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.20014280527830125, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.1306371994316578, + "grad_norm": 0.11735167354345322, + "kept_token_count": 654.4203125, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9799752533435822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020024757739156483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.3790710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87451171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3990, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.1999160096049309, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.12504441142082215, + "grad_norm": 0.10600252449512482, + "kept_token_count": 654.3734375, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9810557067394257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018944281339645385, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.07789611816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1804992675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4000, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.2005310535430908, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.13014456108212472, + "grad_norm": 0.08995647728443146, + "kept_token_count": 654.06875, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9814902305603027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018509764783084394, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.63414306640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.43735961914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4010, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20028143674135207, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.1406739540398121, + "grad_norm": 0.09248088300228119, + "kept_token_count": 654.61875, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.980689013004303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019310978800058366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.57266845703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.00013122558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4020, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.1996585249900818, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.12161458879709244, + "grad_norm": 0.09559828788042068, + "kept_token_count": 655.0625, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9816088974475861, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018391089513897896, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.01475982666017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0451995849609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4030, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.20089830607175826, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.11803177893161773, + "grad_norm": 0.10455549508333206, + "kept_token_count": 653.6734375, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9808283746242523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01917163161560893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.04130096435546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.962255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4040, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2003113329410553, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.11525391265749932, + "grad_norm": 0.09744341671466827, + "kept_token_count": 654.4765625, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9806085824966431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019391422905027867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.44104919433593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4738525390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.1989669382572174, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.12407069131731988, + "grad_norm": 0.10923274606466293, + "kept_token_count": 655.321875, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9823097050189972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017690294422209263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.2470947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1063629150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4060, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2001034915447235, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.13498240560293198, + "grad_norm": 0.10097295045852661, + "kept_token_count": 654.5203125, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9800894081592559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019910589233040808, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.2688201904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9046142578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4070, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.2001594826579094, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.1300273559987545, + "grad_norm": 0.09459424763917923, + "kept_token_count": 654.6515625, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9808555245399475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019144459068775176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.2133987426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6779541015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4080, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2012852907180786, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.1300831615924835, + "grad_norm": 0.13443051278591156, + "kept_token_count": 653.2390625, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.98055859208107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0194414209574461, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.41020660400392, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.66388549804685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4090, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.2013082206249237, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.12759200483560562, + "grad_norm": 0.12248046696186066, + "kept_token_count": 653.4796875, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.980882215499878, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019117768574506044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.98290100097657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6010986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4100, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2002600073814392, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.12380946576595306, + "grad_norm": 0.07547024637460709, + "kept_token_count": 654.74375, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9794119834899903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.020588012970983983, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.4648239135742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2203826904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4110, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.19981636255979537, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.12339811399579048, + "grad_norm": 0.08213220536708832, + "kept_token_count": 654.6921875, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9809152245521545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019084760546684267, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.8249496459961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.58473205566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4120, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19974493384361267, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.11098271384835243, + "grad_norm": 0.07807376235723495, + "kept_token_count": 654.771875, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9812708377838135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01872916929423809, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.85193481445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2776702880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4130, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.19911759793758393, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.13394064530730249, + "grad_norm": 0.08277291059494019, + "kept_token_count": 655.4171875, + "learning_rate": 2.936978999248908e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9810162425041199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018983744829893113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.0274658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9082458496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4140, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.19900876581668853, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.12477331981062889, + "grad_norm": 0.09694111347198486, + "kept_token_count": 655.290625, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9801347553730011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01986525356769562, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.8215759277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.38670349121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4150, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.2001057595014572, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.14009822458028792, + "grad_norm": 0.09595236927270889, + "kept_token_count": 654.48125, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9805854380130767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019414549879729747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.82319183349608, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2115936279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4160, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20085529685020448, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.135874442756176, + "grad_norm": 0.0864219069480896, + "kept_token_count": 653.7140625, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.980328643321991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01967136235907674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.0750305175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.26711120605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4170, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.20121381431818008, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.1257898837327957, + "grad_norm": 0.09164576977491379, + "kept_token_count": 653.5796875, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811172127723694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018882797006517647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.32831115722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.250537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4180, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2004477083683014, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.1232442781329155, + "grad_norm": 0.06508216261863708, + "kept_token_count": 654.4859375, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9806188881397248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019381108693778515, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.2123001098633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5190093994141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4190, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.19983320087194442, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.11663895025849343, + "grad_norm": 0.10336427390575409, + "kept_token_count": 654.6625, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9808956444263458, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019104357250034808, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.09450073242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6523376464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4200, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20045003145933152, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.12459305375814438, + "grad_norm": 0.1044914647936821, + "kept_token_count": 654.1796875, + "learning_rate": 2.933616574579118e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9809620022773743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019038003496825696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.04727478027343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8917938232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4210, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20026336163282393, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.12348791062831879, + "grad_norm": 0.11419620364904404, + "kept_token_count": 654.56875, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.980057156085968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019942847639322282, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.0144515991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.45750427246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4220, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.20003049969673156, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.1282755143940449, + "grad_norm": 0.11930984258651733, + "kept_token_count": 654.89375, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9804880440235137, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019511957839131355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.4996810913086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.97380981445315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4230, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.20064959973096846, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.1338702544569969, + "grad_norm": 0.07729891687631607, + "kept_token_count": 654.14375, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9826513707637787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017348644230514763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.6212631225586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5738250732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4240, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19897702783346177, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.12586216852068902, + "grad_norm": 0.12149038910865784, + "kept_token_count": 655.4640625, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9816400349140167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01835996173322201, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.13702239990235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9604248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4250, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20011419504880906, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.12685397863388062, + "grad_norm": 0.08034864813089371, + "kept_token_count": 654.5359375, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811434209346771, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01885656304657459, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.6779754638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4515716552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4260, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.19986664950847627, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.12466855347156525, + "grad_norm": 0.08942198008298874, + "kept_token_count": 654.6546875, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9815685629844666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01843143440783024, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.8421615600586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6360687255859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4270, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19969926923513412, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.13303281664848327, + "grad_norm": 0.10318374633789062, + "kept_token_count": 654.915625, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9808640539646148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019135932438075542, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.38660278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9850372314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4280, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20047206431627274, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.12951401248574257, + "grad_norm": 0.11350981146097183, + "kept_token_count": 654.3140625, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9800612449645996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019938750378787517, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.41136474609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9857604980469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4290, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19991290867328643, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.1306914657354355, + "grad_norm": 0.09275271743535995, + "kept_token_count": 654.58125, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9814568221569061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018543188087642194, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.595458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9742706298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4300, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.199540351331234, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.1324455589056015, + "grad_norm": 0.08725381642580032, + "kept_token_count": 655.1921875, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9800974130630493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019902583584189414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.07306823730468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.06211547851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4310, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.20048118829727174, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.11863924339413642, + "grad_norm": 0.0999106615781784, + "kept_token_count": 654.1390625, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.980722576379776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01927742436528206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.8288833618164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41590576171876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4320, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.1991927817463875, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.12991444393992424, + "grad_norm": 0.10011347383260727, + "kept_token_count": 655.240625, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9813959658145904, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018604037817567587, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.54423065185546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.93004455566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4330, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.19945454001426696, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.132670146971941, + "grad_norm": 0.11017981171607971, + "kept_token_count": 654.7140625, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9807518780231476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019248125050216912, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.61151275634765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4134490966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4340, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.19941647201776505, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.13169554397463798, + "grad_norm": 0.09276305139064789, + "kept_token_count": 655.0375, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811406552791595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018859352730214596, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.01719665527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1777008056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19965398758649827, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.12813421562314034, + "grad_norm": 0.1144334152340889, + "kept_token_count": 654.728125, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.019, + "mgd_cos_sim": 0.980956757068634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019043240044265986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.7406463623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.85686950683595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4360, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20001115500926972, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.12655918300151825, + "grad_norm": 0.09978364408016205, + "kept_token_count": 654.653125, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9824809789657593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01751903546974063, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.15387725830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6590301513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4370, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19971695244312287, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.13885200545191764, + "grad_norm": 0.11750707775354385, + "kept_token_count": 654.8078125, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9804142057895661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01958578722551465, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.11947631835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5678344726563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4380, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19901614934206008, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.1251350797712803, + "grad_norm": 0.12088119983673096, + "kept_token_count": 655.6109375, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9814557075500489, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018544292077422142, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.51046752929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.68873901367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4390, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.20010424107313157, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.1328434504568577, + "grad_norm": 0.09949152916669846, + "kept_token_count": 654.771875, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9805524051189423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01944758351892233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.34251708984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.08128356933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4400, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.2002897694706917, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.13323383927345275, + "grad_norm": 0.10190374404191971, + "kept_token_count": 654.6265625, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9809891581535339, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019010864198207855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.0145050048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7340515136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.19991349726915358, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.1316762700676918, + "grad_norm": 0.09783521294593811, + "kept_token_count": 654.9921875, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9810517728328705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018948230426758528, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.1083251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.46860656738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4420, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.20019017457962035, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.13289180397987366, + "grad_norm": 0.10559666901826859, + "kept_token_count": 654.45625, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9816634595394135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018336526304483413, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.1345977783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.04877014160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4430, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20091105699539186, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.12504695132374763, + "grad_norm": 0.07265470921993256, + "kept_token_count": 654.0984375, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9821934640407562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01780654089525342, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.50055694580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3025268554687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4440, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.20052022933959962, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.11731774806976318, + "grad_norm": 0.08522592484951019, + "kept_token_count": 654.15625, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9813845038414002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01861548749729991, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.18500213623048, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.28020629882815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.20031708776950835, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.11819032952189445, + "grad_norm": 0.09590034186840057, + "kept_token_count": 653.9984375, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9819132864475251, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018086705543100833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.6281524658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07207946777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4460, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.19919707328081132, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.12110223099589348, + "grad_norm": 0.08207959681749344, + "kept_token_count": 655.2234375, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9823835730552674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01761643188074231, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.73475036621093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0507537841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4470, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.20022763460874557, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.12607573866844177, + "grad_norm": 0.09216311573982239, + "kept_token_count": 654.421875, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.981486302614212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018513705767691135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.43891296386718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.79961853027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4480, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20002794563770293, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.11448767855763435, + "grad_norm": 0.12115751951932907, + "kept_token_count": 654.409375, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9814483404159546, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018551670387387274, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.6064208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1580108642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4490, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.19966356605291366, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.1353042371571064, + "grad_norm": 0.10110880434513092, + "kept_token_count": 654.7609375, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9816583395004272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01834166208282113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.08645629882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.96552734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4500, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20020826607942582, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.11051755100488662, + "grad_norm": 0.08737824857234955, + "kept_token_count": 654.521875, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9812254309654236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018774560000747443, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.2511444091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3801727294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4510, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2006465747952461, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.11398685127496719, + "grad_norm": 0.11077536642551422, + "kept_token_count": 654.00625, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9814094603061676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018590531218796967, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.96695861816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8080535888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4520, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.1998330146074295, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.11563069745898247, + "grad_norm": 0.10060601681470871, + "kept_token_count": 654.5515625, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.982327800989151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017672192584723235, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.43622283935548, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3511505126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.1985391706228256, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.1167817659676075, + "grad_norm": 0.10045114159584045, + "kept_token_count": 655.7578125, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9801861345767975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019813866633921862, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.43228759765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8717864990234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4540, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.20090820640325546, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.11465739160776138, + "grad_norm": 0.08762213587760925, + "kept_token_count": 653.8875, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9814864218235015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018513576406985522, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.21846466064454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.47699584960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4550, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.19905334413051606, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.11676691398024559, + "grad_norm": 0.08549141138792038, + "kept_token_count": 655.2828125, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9831948101520538, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016805186308920382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.15824127197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5534301757813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4560, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.1996058776974678, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.11634154915809632, + "grad_norm": 0.106780506670475, + "kept_token_count": 654.909375, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9822031319141388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017796868085861207, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.75703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.05633544921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4570, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.1985940530896187, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.12352093905210496, + "grad_norm": 0.08042877167463303, + "kept_token_count": 655.5015625, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9819579899311066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018042025715112688, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.04761199951173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35439453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4580, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.1999767765402794, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.11856717467308045, + "grad_norm": 0.09649978578090668, + "kept_token_count": 654.7328125, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9810447037220001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01895528454333544, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.3187484741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.48948974609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4590, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.20107646584510802, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.11414771154522896, + "grad_norm": 0.09333094954490662, + "kept_token_count": 653.5140625, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9825401544570923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017459851689636707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.2774642944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2973358154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4600, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.20044954419136046, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.13103836476802827, + "grad_norm": 0.07704227417707443, + "kept_token_count": 654.2203125, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9820124626159668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017987542878836395, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.46592559814454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.54001770019534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4610, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20098492354154587, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.11952729597687721, + "grad_norm": 0.1068253144621849, + "kept_token_count": 653.6453125, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9821689844131469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0178310121409595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.03833465576173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4624969482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4620, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.2004472240805626, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.12287515327334404, + "grad_norm": 0.084188312292099, + "kept_token_count": 654.346875, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9822218000888825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017778191715478897, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.7349105834961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7732391357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4630, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19909707009792327, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.11605842784047127, + "grad_norm": 0.10020709782838821, + "kept_token_count": 655.434375, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9825480341911316, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01745196096599102, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.81401519775392, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5299102783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4640, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19982649981975556, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.11548011377453804, + "grad_norm": 0.08184072375297546, + "kept_token_count": 654.6640625, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811170518398284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018882950581610203, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 196.3694595336914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.31725158691404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20027344971895217, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.12228955924510956, + "grad_norm": 0.1155870258808136, + "kept_token_count": 654.58125, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.98235382437706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01764617282897234, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.94911193847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.14125671386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4660, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.19922355115413665, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.13763300254940986, + "grad_norm": 0.07929503172636032, + "kept_token_count": 655.1875, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830010890960693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016998902428895235, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.34276123046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0774810791016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4670, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19955069273710252, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.13446256816387175, + "grad_norm": 0.11170198768377304, + "kept_token_count": 655.125, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9820649147033691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017935086227953433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.9180114746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9175323486328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4680, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20049460530281066, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.13207057267427444, + "grad_norm": 0.11153429746627808, + "kept_token_count": 654.2484375, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9822415471076965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01775844907388091, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.71243896484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2755523681641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4690, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.200641730427742, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.12620218396186828, + "grad_norm": 0.10689966380596161, + "kept_token_count": 653.9546875, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811062633991241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01889373119920492, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 199.2195297241211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2409332275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4700, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20050009787082673, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.11346795037388802, + "grad_norm": 0.10512863844633102, + "kept_token_count": 654.2484375, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9824141025543213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01758591216057539, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.58912200927733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4848327636719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4710, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.20030216723680497, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.1162404403090477, + "grad_norm": 0.07659633457660675, + "kept_token_count": 654.7515625, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9814956784248352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018504324182868005, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.05321350097657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4325805664063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4720, + "valid_token_count": 818.75 + }, + { + "actual_token_mask_ratio": 0.19998277872800826, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.12057065516710282, + "grad_norm": 0.07706847786903381, + "kept_token_count": 654.4359375, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9824213266372681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017578688357025384, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.87783813476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2849853515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4730, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.2004037857055664, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.12241809219121932, + "grad_norm": 0.10438975691795349, + "kept_token_count": 654.2578125, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9832106292247772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01678937915712595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 206.32692108154296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4120758056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4740, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2015633761882782, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.13158896341919898, + "grad_norm": 0.09237968176603317, + "kept_token_count": 653.5078125, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9818941116333008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01810589348897338, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.91888427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2334503173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4750, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.2011793449521065, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.12590473219752313, + "grad_norm": 0.08164013177156448, + "kept_token_count": 653.5109375, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9826540172100067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01734598269686103, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.58767395019532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0685638427734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.2004605397582054, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.12487789765000343, + "grad_norm": 0.14711295068264008, + "kept_token_count": 654.090625, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9818357229232788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018164288625121115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.22043151855468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30860900878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4770, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20000616759061812, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.13366350010037423, + "grad_norm": 0.09511634707450867, + "kept_token_count": 654.6734375, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9811043381690979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018895654287189246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.48064270019532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3480926513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4780, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20074650645256042, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.13100979775190352, + "grad_norm": 0.08846239745616913, + "kept_token_count": 654.0875, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.982507586479187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017492411099374293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 209.77532196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.44341735839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4790, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19926706403493882, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.11356252580881118, + "grad_norm": 0.08980189263820648, + "kept_token_count": 655.4984375, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9832966566085816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016703351214528084, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.7705337524414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5561096191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4800, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.2002960056066513, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.1266588695347309, + "grad_norm": 0.10488148778676987, + "kept_token_count": 654.4828125, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9833714187145233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016628572437912225, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.2036911010742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0021606445313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4810, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.200175042450428, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.12683279067277908, + "grad_norm": 0.08760494738817215, + "kept_token_count": 654.525, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9828643321990966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017135663609951734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.92178955078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6126373291016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4820, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19999865889549256, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.12779465913772584, + "grad_norm": 0.09361991286277771, + "kept_token_count": 654.7734375, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9825218260288239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017478164564818145, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.921923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.40198974609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4830, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.2000703975558281, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.12719825133681298, + "grad_norm": 0.09103965759277344, + "kept_token_count": 654.58125, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9827529966831208, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017246992234140633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.9305618286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5903594970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4840, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19969281703233718, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.12004832848906517, + "grad_norm": 0.07013031095266342, + "kept_token_count": 654.9640625, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9834629833698273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01653702510520816, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.01179962158204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0884124755859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4850, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19958774000406265, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.13587753251194953, + "grad_norm": 0.07485757768154144, + "kept_token_count": 655.05625, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9830971658229828, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016902835201472045, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.61485900878907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84454650878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4860, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20046427845954895, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.13517037630081177, + "grad_norm": 0.08180458843708038, + "kept_token_count": 654.2515625, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9810269355773926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018973053712397814, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.44920043945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.870751953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4870, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20095474123954774, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.13330647200345994, + "grad_norm": 0.11261311918497086, + "kept_token_count": 653.7109375, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9820946514606476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01790534518659115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.52549896240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.75279846191404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4880, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19988831281661987, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.11430377587676048, + "grad_norm": 0.07706067711114883, + "kept_token_count": 654.6125, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9827904462814331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017209562193602323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.9173843383789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.87457580566405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4890, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19913882166147232, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.12221190184354783, + "grad_norm": 0.07605544477701187, + "kept_token_count": 655.421875, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9828922271728515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017107768543064596, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.55581512451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0663238525391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4900, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.200615856051445, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.13058193251490594, + "grad_norm": 0.13332432508468628, + "kept_token_count": 653.9171875, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9820608675479889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017939143255352973, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.39080810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0375457763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4910, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.19935811311006546, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.13374375998973848, + "grad_norm": 0.07315289229154587, + "kept_token_count": 654.909375, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9807218194007874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01927817752584815, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.77662658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.62701110839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4920, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.19989222139120102, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.12670763731002807, + "grad_norm": 0.09597249329090118, + "kept_token_count": 654.75625, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.984183257818222, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01581673314794898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.40889739990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.12193908691404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4930, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.1988646626472473, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.1224076084792614, + "grad_norm": 0.09213482588529587, + "kept_token_count": 655.6890625, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9822210967540741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017778891324996948, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.17987060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1544952392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4940, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.1993767499923706, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.12128799483180046, + "grad_norm": 0.12715008854866028, + "kept_token_count": 655.1015625, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9828140139579773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01718599395826459, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.13313903808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.20179138183596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.1994193896651268, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.12468064650893211, + "grad_norm": 0.0753793865442276, + "kept_token_count": 655.2703125, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9827276587486267, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01727233864367008, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.6292694091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8201202392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4960, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.19895011186599731, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.13392048254609107, + "grad_norm": 0.0787530466914177, + "kept_token_count": 655.4578125, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9824237287044525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017576271761208774, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.38246002197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.78175964355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4970, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2000449225306511, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.13422342166304588, + "grad_norm": 0.08161982893943787, + "kept_token_count": 654.4890625, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9828998208045959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017100169602781535, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.1426986694336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9135040283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2008614346385002, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.1265537291765213, + "grad_norm": 0.11324897408485413, + "kept_token_count": 653.8359375, + "learning_rate": 2.890424516818774e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9829701542854309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017029840871691702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.7527587890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.30726013183596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4990, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19997314512729644, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.12390861362218857, + "grad_norm": 0.08088042587041855, + "kept_token_count": 654.871875, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9832401096820831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016759898606687786, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.75179901123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.810888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5000, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.20012577176094054, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.126387257874012, + "grad_norm": 0.09140130132436752, + "kept_token_count": 654.640625, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9835628926753998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016437098011374475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.45856628417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1656829833984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5010, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19941382557153703, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.13252767994999887, + "grad_norm": 0.09356366097927094, + "kept_token_count": 655.084375, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9826890289783478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017310956865549086, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.92157745361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.88983154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20048271864652634, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.1252443790435791, + "grad_norm": 0.08009669929742813, + "kept_token_count": 654.453125, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.982859629392624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01714037321507931, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.08365020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.07472839355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5030, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.19980309754610062, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.12489562705159188, + "grad_norm": 0.05759464576840401, + "kept_token_count": 654.928125, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.983073627948761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016926380712538956, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.61907043457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.04730529785155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5040, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.19902615547180175, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.12033074423670768, + "grad_norm": 0.09045621007680893, + "kept_token_count": 655.6796875, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.983452957868576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016547045111656188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.8306869506836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2201751708984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5050, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.1997843414545059, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.12167467027902604, + "grad_norm": 0.10477624833583832, + "kept_token_count": 654.6359375, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9841586172580719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015841390751302243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.14251708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.367529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5060, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.20036894977092742, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.12549338564276696, + "grad_norm": 0.09450340270996094, + "kept_token_count": 654.4984375, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9832863211631775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016713673900812864, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.1704559326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9027038574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5070, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19987103939056397, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.1280483603477478, + "grad_norm": 0.10348167270421982, + "kept_token_count": 654.878125, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9837018728256226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016298141796141863, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.9332290649414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3454193115234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19943735599517823, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.12826571762561798, + "grad_norm": 0.10708349943161011, + "kept_token_count": 655.015625, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9832125782966614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016787410993129016, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.613037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.97755126953126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5090, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19968530386686326, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.12944849133491515, + "grad_norm": 0.06689593940973282, + "kept_token_count": 654.6609375, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9833701074123382, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016629879642277957, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.77205505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.56585693359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.19986691176891327, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.12481077611446381, + "grad_norm": 0.06906572729349136, + "kept_token_count": 654.7265625, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9828351974487305, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01716481726616621, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 212.09320526123048, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.53432006835936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2000904232263565, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.13053532615303992, + "grad_norm": 0.09008877724409103, + "kept_token_count": 654.4625, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9824704647064209, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01752954861149192, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.42725982666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.26576232910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5120, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.20062539726495743, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.12394530922174454, + "grad_norm": 0.08934232592582703, + "kept_token_count": 654.26875, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9837008655071259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01629912843927741, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.19872131347657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3294219970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5130, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.20055946111679077, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.14592393189668657, + "grad_norm": 0.0850830003619194, + "kept_token_count": 654.2, + "learning_rate": 2.880929503756906e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9809912741184235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.019008718617260455, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.93468017578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2799377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5140, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2001267746090889, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.13050466775894165, + "grad_norm": 0.0775032490491867, + "kept_token_count": 654.2625, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9824588000774384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017541184928268193, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.3401870727539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8640075683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5150, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.1993450865149498, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.1336825855076313, + "grad_norm": 0.09230378270149231, + "kept_token_count": 655.3671875, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9817492365837097, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.018250775802880526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.7551742553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10352783203126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5160, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.20128220170736313, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.12669872716069222, + "grad_norm": 0.07658012211322784, + "kept_token_count": 653.578125, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9838685393333435, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016131462715566158, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.33937377929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.71576232910155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5170, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.2007072314620018, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.13213590160012245, + "grad_norm": 0.07719319313764572, + "kept_token_count": 654.0328125, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9828336536884308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01716635311022401, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.20528106689454, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.77735900878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5180, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19886421263217927, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.13430483862757683, + "grad_norm": 0.09385868906974792, + "kept_token_count": 655.584375, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834111332893372, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01658885581418872, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.58578186035157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8411102294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5190, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.19965892434120178, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.120171557366848, + "grad_norm": 0.080825574696064, + "kept_token_count": 654.95, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830199599266052, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016980032622814178, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.53223571777343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.56090393066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5200, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20065479576587678, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.12402929812669754, + "grad_norm": 0.09589826315641403, + "kept_token_count": 653.94375, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9838425815105438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01615740917623043, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.17760467529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1182922363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5210, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.19968436062335967, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.12089252471923828, + "grad_norm": 0.0757191926240921, + "kept_token_count": 654.8953125, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9840861797332764, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015913831628859045, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.9172805786133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55833129882814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5220, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20016024708747865, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.12720787450671195, + "grad_norm": 0.07169414311647415, + "kept_token_count": 654.4234375, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9829548835754395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017045110277831555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.87470397949218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4806762695313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5230, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2005972683429718, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.11994469985365867, + "grad_norm": 0.10652069002389908, + "kept_token_count": 654.115625, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9826295852661133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01737041659653187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.50473175048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7403839111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5240, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19955670535564424, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.12475561425089836, + "grad_norm": 0.08308237046003342, + "kept_token_count": 655.284375, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9838895559310913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016110459808260202, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.81773376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3636779785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5250, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.20064249038696289, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.13721410408616067, + "grad_norm": 0.08656220883131027, + "kept_token_count": 654.1515625, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.98206627368927, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017933720722794533, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.3440689086914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4674926757813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5260, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20043190568685532, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.12606335654854775, + "grad_norm": 0.10907690227031708, + "kept_token_count": 654.46875, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9834810137748718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016518985200673342, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.5414047241211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7595184326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5270, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.1996765300631523, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.12680638507008551, + "grad_norm": 0.0864226371049881, + "kept_token_count": 654.646875, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9827929556369781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01720704771578312, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.21497650146483, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8290679931641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5280, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.198737470805645, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.1293739140033722, + "grad_norm": 0.06561031192541122, + "kept_token_count": 655.5890625, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834217429161072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016578255407512187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.12775115966798, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.499560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5290, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20045503973960876, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.1301554821431637, + "grad_norm": 0.08914142102003098, + "kept_token_count": 654.4578125, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9835676610469818, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01643233709037304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.66370697021483, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0897918701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5300, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.20021245777606964, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.12358860969543457, + "grad_norm": 0.08464820683002472, + "kept_token_count": 654.3234375, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9835827589035034, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016417225636541844, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.1958969116211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6584197998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5310, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20001524686813354, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.13326261714100837, + "grad_norm": 0.09222403913736343, + "kept_token_count": 654.8328125, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9826268374919891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017373147420585155, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.37089996337892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8308502197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5320, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.20036228001117706, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.12241432964801788, + "grad_norm": 0.09264153242111206, + "kept_token_count": 654.478125, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834131062030792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016586882807314397, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.10950622558593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2285522460937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2000378727912903, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.12136455178260804, + "grad_norm": 0.08549241721630096, + "kept_token_count": 654.571875, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834320187568665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016567984689027072, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.22882843017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1674743652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5340, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19957218766212464, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.11431717053055764, + "grad_norm": 0.08643832802772522, + "kept_token_count": 655.1015625, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9829721271991729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01702786926180124, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.4612060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.199560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5350, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.1996144399046898, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.12139037698507309, + "grad_norm": 0.11028317362070084, + "kept_token_count": 654.825, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.017, + "mgd_cos_sim": 0.982973700761795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01702629551291466, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.35497589111327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2668975830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5360, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.199686898291111, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.12342562377452851, + "grad_norm": 0.08201641589403152, + "kept_token_count": 654.775, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9825709700584412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01742903348058462, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.9024627685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1711853027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5370, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.19985398799180984, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.13672752007842065, + "grad_norm": 0.06946005672216415, + "kept_token_count": 654.64375, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.016, + "mgd_cos_sim": 0.983958387374878, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016041609831154345, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.37699279785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1776153564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2004215195775032, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.13450254797935485, + "grad_norm": 0.08871250599622726, + "kept_token_count": 654.146875, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9833268642425537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0166731265373528, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.02592010498046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19943317025899887, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.13351638689637185, + "grad_norm": 0.08505208790302277, + "kept_token_count": 655.3796875, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9837768018245697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016223182249814273, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.57274627685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3012664794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5400, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.20042191445827484, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.11677023321390152, + "grad_norm": 0.08749087154865265, + "kept_token_count": 654.2609375, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9828728914260865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017127107642591, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.4430679321289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.42649230957034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20109007209539415, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.12596061900258065, + "grad_norm": 0.07492611557245255, + "kept_token_count": 653.6046875, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.984079784154892, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015920223947614432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.77487030029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19312438964846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5420, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19954753071069717, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.11567806378006935, + "grad_norm": 0.07578316330909729, + "kept_token_count": 654.9453125, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834064781665802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016593511588871478, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.22437744140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6289855957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5430, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19931508749723434, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.11607774943113328, + "grad_norm": 0.0907062292098999, + "kept_token_count": 655.1640625, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9843767046928406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015623294189572335, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.2108917236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5444854736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5440, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20034307688474656, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.12302327156066895, + "grad_norm": 0.11368761211633682, + "kept_token_count": 654.3, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9826779127120971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017322077136486768, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.68096313476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.99939880371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5450, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20002381950616838, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.1225696012377739, + "grad_norm": 0.09099435806274414, + "kept_token_count": 654.6875, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9831615507602691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016838465072214603, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.19267883300782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4386749267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5460, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.1995517149567604, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.13095950409770013, + "grad_norm": 0.08435476571321487, + "kept_token_count": 654.8953125, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9826217472553254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01737825954332948, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.25287933349608, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8288818359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5470, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20001961290836334, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.1256137065589428, + "grad_norm": 0.09240879863500595, + "kept_token_count": 654.61875, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9827784478664399, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017221543844789265, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.63357849121093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19928283691405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19974254816770554, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.1268589586019516, + "grad_norm": 0.0779280886054039, + "kept_token_count": 654.928125, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9836270570755005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01637294692918658, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.7259262084961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.65732421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5490, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20005736500024796, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.13332032710313796, + "grad_norm": 0.07598448544740677, + "kept_token_count": 654.71875, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9826353192329407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0173646854236722, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.912939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.5520416259766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.19929393231868744, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.12440094575285912, + "grad_norm": 0.1393735259771347, + "kept_token_count": 655.075, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9834196507930756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01658033886924386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.43014831542968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9357513427734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5510, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.20102381259202956, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.12266491129994392, + "grad_norm": 0.08814798295497894, + "kept_token_count": 653.6890625, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9836535930633545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01634641457349062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.16968841552733, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.105322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5520, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.2001951441168785, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.1216631680727005, + "grad_norm": 0.08486765623092651, + "kept_token_count": 654.3921875, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9836367666721344, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016363224294036628, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.99237518310548, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6691467285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5530, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19943049252033235, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.1342901460826397, + "grad_norm": 0.0745818093419075, + "kept_token_count": 655.2734375, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830394923686981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01696051238104701, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.93787078857423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1917724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5540, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19934877008199692, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.13117164447903634, + "grad_norm": 0.10588282346725464, + "kept_token_count": 654.9890625, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9840057075023652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015994297899305822, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.18993072509767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0653564453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5550, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20054731220006944, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.13837194964289665, + "grad_norm": 0.09113480150699615, + "kept_token_count": 654.471875, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9826757788658143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017324214056134223, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.7460723876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5586853027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5560, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.19934413582086563, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.1134135402739048, + "grad_norm": 0.0921161100268364, + "kept_token_count": 655.4390625, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9841041505336762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01589584704488516, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.8535354614258, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.1721527099609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5570, + "valid_token_count": 818.6265625 + }, + { + "actual_token_mask_ratio": 0.19988557398319245, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.13306825533509253, + "grad_norm": 0.07565665990114212, + "kept_token_count": 655.0046875, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9831062376499176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016893763560801744, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.93951873779298, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1000518798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5580, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20036921203136443, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.12527931109070778, + "grad_norm": 0.08170486986637115, + "kept_token_count": 654.2328125, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9833825290203094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016617469675838947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.0975555419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.85142211914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20038450062274932, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.12064014002680779, + "grad_norm": 0.07637853920459747, + "kept_token_count": 654.5078125, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9836619794368744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016338008269667626, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.43674011230468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6462738037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5600, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.20010843873023987, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.12428107708692551, + "grad_norm": 0.09158003330230713, + "kept_token_count": 654.6828125, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9827640950679779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017235902976244687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.7459503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9333923339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2001202404499054, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.11940102130174637, + "grad_norm": 0.07859064638614655, + "kept_token_count": 654.3640625, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9843726754188538, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015627321135252715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.29954833984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14792785644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5620, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.200753653049469, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.12268741577863693, + "grad_norm": 0.0827094316482544, + "kept_token_count": 654.0328125, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9839486837387085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01605132585391402, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.7018829345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9853454589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5630, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.19953446239233016, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.12552816048264503, + "grad_norm": 0.07687448710203171, + "kept_token_count": 654.9609375, + "learning_rate": 2.846568190920773e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9840138137340546, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015986185986548662, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.06875762939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1266632080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5640, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19976794570684434, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.13441549614071846, + "grad_norm": 0.0804712325334549, + "kept_token_count": 655.0296875, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9837099730968475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016290019452571868, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.6464050292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7886657714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5650, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.19967978447675705, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.12037076503038406, + "grad_norm": 0.07813974469900131, + "kept_token_count": 654.721875, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.982530266046524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017469719424843787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 227.7953109741211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0048522949219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5660, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.19961059093475342, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.13040651679039, + "grad_norm": 0.10400091856718063, + "kept_token_count": 654.765625, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9839952111244201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016004797723144292, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.16297607421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3374053955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5670, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.19973499625921248, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.11808059513568878, + "grad_norm": 0.08494824171066284, + "kept_token_count": 654.825, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9831326425075531, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016867364570498467, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.3936996459961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.94044799804686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5680, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20097641944885253, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.12701571360230446, + "grad_norm": 0.08447425067424774, + "kept_token_count": 653.696875, + "learning_rate": 2.842905307516495e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830376327037811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016962359379976987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.45525207519532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8909851074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20031383037567138, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.11297912448644638, + "grad_norm": 0.08599788695573807, + "kept_token_count": 654.434375, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9840563774108887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015943621378391983, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.9389190673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2903625488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5700, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19966231137514115, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.12731879502534865, + "grad_norm": 0.08983884006738663, + "kept_token_count": 655.1390625, + "learning_rate": 2.841428725613801e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9829573929309845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01704259468242526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.16780853271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9117736816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5710, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.20028368830680848, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.1253532201051712, + "grad_norm": 0.07396251708269119, + "kept_token_count": 654.4265625, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9840330898761749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015966907143592834, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.8675765991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5051025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.20008741915225983, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.13182082623243332, + "grad_norm": 0.0823090448975563, + "kept_token_count": 654.378125, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830170512199402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016982949711382388, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.3362060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3284423828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5730, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20060105472803116, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.12607187405228615, + "grad_norm": 0.07961340248584747, + "kept_token_count": 653.9984375, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9832297146320343, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016770283784717323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.52867889404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0999816894531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.1997488871216774, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.11972498819231987, + "grad_norm": 0.06687984615564346, + "kept_token_count": 654.95, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.98391193151474, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016088060662150384, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.23377380371093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5750, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19941600561141967, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.11770730316638947, + "grad_norm": 0.08525505661964417, + "kept_token_count": 654.8703125, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.984463894367218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015536102838814259, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.9088333129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4938049316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5760, + "valid_token_count": 817.9921875 + }, + { + "actual_token_mask_ratio": 0.20002882033586503, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.11944987326860428, + "grad_norm": 0.07885296642780304, + "kept_token_count": 654.4703125, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850106418132782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014989361353218555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.95269012451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8161193847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5770, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19977938383817673, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.132228584587574, + "grad_norm": 0.06088605895638466, + "kept_token_count": 654.71875, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9842222034931183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01577779818326235, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.17474365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66583557128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5780, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1998150572180748, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.12613585293293, + "grad_norm": 0.07498899847269058, + "kept_token_count": 654.9859375, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848282814025879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015171722788363696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.85727233886718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6374969482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5790, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.19954300820827484, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.12197979614138603, + "grad_norm": 0.10820610076189041, + "kept_token_count": 654.8640625, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9836739659309387, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01632603695616126, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.7414337158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1392822265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5800, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20030635446310044, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.12035120725631714, + "grad_norm": 0.1332385092973709, + "kept_token_count": 654.2453125, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845227360725403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015477251820266246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.90296783447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.36785888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5810, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19977272003889085, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.11374101191759109, + "grad_norm": 0.11827508360147476, + "kept_token_count": 654.7140625, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9839389741420745, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016061018314212562, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.07528839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48163452148435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5820, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19911163449287414, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.13061708211898804, + "grad_norm": 0.06380221247673035, + "kept_token_count": 655.1921875, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9840732276439667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015926765836775304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.39971008300782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.29803161621095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5830, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19983735978603362, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.11410021707415581, + "grad_norm": 0.09491689503192902, + "kept_token_count": 654.7578125, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9851010024547577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014898986835032702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.82099304199218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.28350219726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5840, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.2006969928741455, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.12326682731509209, + "grad_norm": 0.09160589426755905, + "kept_token_count": 653.91875, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9833468794822693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016653122659772635, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.5404510498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.66802978515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5850, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.19988512396812438, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.12621696442365646, + "grad_norm": 0.075124092400074, + "kept_token_count": 654.45625, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847149550914764, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01528505515307188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.73929748535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.53106079101565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5860, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.19941439777612685, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.1182808093726635, + "grad_norm": 0.09777872264385223, + "kept_token_count": 654.9953125, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9828116953372955, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.017188295256346464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.60215911865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.415673828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5870, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19959367513656617, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.12803250774741173, + "grad_norm": 0.08146882057189941, + "kept_token_count": 654.8234375, + "learning_rate": 2.828615273381551e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9830446183681488, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01695539029315114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.80252838134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.32005310058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5880, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.1999952480196953, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.11859463974833488, + "grad_norm": 0.08865687996149063, + "kept_token_count": 654.8078125, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9842996954917907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015700313076376916, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.04230194091798, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.52626342773436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5890, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19928414672613143, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.1154718279838562, + "grad_norm": 0.0630941390991211, + "kept_token_count": 655.225, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9843780398368835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015621956903487444, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.0305191040039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9785675048828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5900, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20001047998666763, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.1175637312233448, + "grad_norm": 0.0752042606472969, + "kept_token_count": 654.3828125, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9851212441921234, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014878749195486308, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.6035140991211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.32911376953126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.19999023973941804, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.12922904044389724, + "grad_norm": 0.07918442785739899, + "kept_token_count": 654.696875, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9833634972572327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016636494267731904, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.32341918945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6346435546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5920, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.19958049058914185, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.1231154128909111, + "grad_norm": 0.09650430083274841, + "kept_token_count": 655.08125, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9832134306430816, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01678656842559576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.88773040771486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.84180908203126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5930, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19986730962991714, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.12105871066451072, + "grad_norm": 0.08991938829421997, + "kept_token_count": 654.7109375, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9835876941680908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01641230322420597, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.3338409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6068542480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5940, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20014672428369523, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.1313985764980316, + "grad_norm": 0.0974792018532753, + "kept_token_count": 654.5859375, + "learning_rate": 2.823203278868685e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9839677393436432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016032262332737447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.21160888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84727783203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5950, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20023249238729476, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.12279110625386239, + "grad_norm": 0.062302447855472565, + "kept_token_count": 654.65625, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845095753669739, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01549041261896491, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.76478729248046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.91651611328126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5960, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.1988161712884903, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.11976798400282859, + "grad_norm": 0.05723627656698227, + "kept_token_count": 655.5671875, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9853039443492889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014696057047694921, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.0858596801758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6087188720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5970, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19988627433776857, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.11067611426115036, + "grad_norm": 0.0815173014998436, + "kept_token_count": 654.9265625, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847488105297089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015251196455210447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.03398742675782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5918029785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5980, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.2004046246409416, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.1307870104908943, + "grad_norm": 0.07256358116865158, + "kept_token_count": 654.440625, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854570508003235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014542951062321664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.6260208129883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4024261474609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5990, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19986193031072616, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.12718222960829734, + "grad_norm": 0.10379436612129211, + "kept_token_count": 655.00625, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9842721462249756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01572785321623087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.52705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3769073486328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6000, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.1990807071328163, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.1253121368587017, + "grad_norm": 0.07804440706968307, + "kept_token_count": 655.1328125, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850324869155884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014967522025108338, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.2500762939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7148468017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6010, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.19974322319030763, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.13453874960541726, + "grad_norm": 0.08461353927850723, + "kept_token_count": 654.828125, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845614016056061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015438616834580898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.77487182617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.74139404296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6020, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20026227384805678, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.1325504772365093, + "grad_norm": 0.09287836402654648, + "kept_token_count": 654.29375, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9835259914398193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016474000364542007, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.20944061279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.840771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6030, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.2007434695959091, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.1182126261293888, + "grad_norm": 0.08743858337402344, + "kept_token_count": 653.6875, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9846495568752289, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015350434836000204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.2961456298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.32762145996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6040, + "valid_token_count": 817.86875 + }, + { + "actual_token_mask_ratio": 0.1999087452888489, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.12699621245265008, + "grad_norm": 0.07159389555454254, + "kept_token_count": 654.7953125, + "learning_rate": 2.815335275225068e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9839734733104706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01602652287110686, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.34809112548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.06435241699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6050, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19998642057180405, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.1267765522003174, + "grad_norm": 0.10354438424110413, + "kept_token_count": 654.7484375, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9837404727935791, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016259533911943437, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.90340270996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1075408935547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6060, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20057989060878753, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.12645454928278924, + "grad_norm": 0.0764758288860321, + "kept_token_count": 654.36875, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9835247159004211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016475296951830386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.2615203857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7776733398438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6070, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.19928274154663086, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.12985387742519378, + "grad_norm": 0.09978658705949783, + "kept_token_count": 654.95, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9841622591018677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015837734658271076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.60867767333986, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.905615234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6080, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.20025641918182374, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.13269202262163163, + "grad_norm": 0.08685295283794403, + "kept_token_count": 654.2625, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9840662956237793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015933708008378743, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.4530258178711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1311065673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6090, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19940183609724044, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.12320265844464302, + "grad_norm": 0.07311428338289261, + "kept_token_count": 655.2671875, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848441660404206, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01515584196895361, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.2903335571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0106231689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6100, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.20026973336935044, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.12177018523216247, + "grad_norm": 0.0658799484372139, + "kept_token_count": 654.9109375, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9838243842124939, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01617561196908355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.1491683959961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1059539794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6110, + "valid_token_count": 818.9234375 + }, + { + "actual_token_mask_ratio": 0.19992734491825104, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.13314545899629593, + "grad_norm": 0.08507490158081055, + "kept_token_count": 654.740625, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844518840312958, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015548129193484783, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.4333709716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6525512695313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6120, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2004357323050499, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.12773193269968033, + "grad_norm": 0.09700700640678406, + "kept_token_count": 654.0578125, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9843434453010559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015656549111008645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.4043365478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8827209472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6130, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.20042714625597, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.1243012823164463, + "grad_norm": 0.08663024753332138, + "kept_token_count": 654.1140625, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9852997779846191, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014700230862945318, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.70008239746093, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7230194091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6140, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20109756141901017, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.1126336209475994, + "grad_norm": 0.08536585420370102, + "kept_token_count": 653.74375, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847992599010468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015200743451714516, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.15692749023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.71856384277345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6150, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20106070786714553, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.1222439855337143, + "grad_norm": 0.06723256409168243, + "kept_token_count": 653.603125, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9831578493118286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01684215757995844, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.2713836669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.35264892578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20004826188087463, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.11868465840816497, + "grad_norm": 0.06983395665884018, + "kept_token_count": 654.559375, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848366737365722, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015163325425237417, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.6145492553711, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2788452148437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6170, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19935175627470017, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.12183300331234932, + "grad_norm": 0.09755851328372955, + "kept_token_count": 655.2609375, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.983199393749237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01680060736835003, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.47187042236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.86913146972654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6180, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20012375563383103, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.12052604332566261, + "grad_norm": 0.07303497195243835, + "kept_token_count": 654.415625, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9848734736442566, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015126525983214378, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.93990478515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.88359985351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6190, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19942513406276702, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.12349980771541595, + "grad_norm": 0.08534728735685349, + "kept_token_count": 655.1578125, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847235083580017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01527650011703372, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.82675170898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9177276611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6200, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.20011489391326903, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.12427404597401619, + "grad_norm": 0.07659939676523209, + "kept_token_count": 654.4109375, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9844391763210296, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015560827683657408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.46995697021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3641693115234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6210, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.20013364106416703, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.11563060730695725, + "grad_norm": 0.06529368460178375, + "kept_token_count": 654.4859375, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9841542363166809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015845749992877245, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.80558319091796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7334899902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6220, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.1999726712703705, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.1238450676202774, + "grad_norm": 0.09448780119419098, + "kept_token_count": 654.746875, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9835774302482605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016422587167471647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.3851028442383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0590881347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20014859437942506, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.13597528263926506, + "grad_norm": 0.06839918345212936, + "kept_token_count": 654.2578125, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9855010986328125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014498911611735821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.16565856933593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9498260498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6240, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20048487633466722, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.12399910688400269, + "grad_norm": 0.07631922513246536, + "kept_token_count": 654.325, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9855952680110931, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014404732827097177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.63121337890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0871551513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6250, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.1997786596417427, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.12478441447019577, + "grad_norm": 0.0854945033788681, + "kept_token_count": 654.878125, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.984482717514038, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015517288818955421, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.8041778564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6265899658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6260, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19984499365091324, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.11355591043829918, + "grad_norm": 0.08440830558538437, + "kept_token_count": 654.6375, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9855810105800629, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014418991934508085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.05819396972657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0246307373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6270, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2001452088356018, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.11017717644572259, + "grad_norm": 0.12265273928642273, + "kept_token_count": 654.2984375, + "learning_rate": 2.796633941131189e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850255012512207, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01497450303286314, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.4320098876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.72129211425784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6280, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20040292888879777, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.1303618937730789, + "grad_norm": 0.08323264122009277, + "kept_token_count": 654.55, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845015525817871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015498445462435484, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.3965438842773, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.08600158691405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6290, + "valid_token_count": 818.596875 + }, + { + "actual_token_mask_ratio": 0.1999015673995018, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.12184631004929543, + "grad_norm": 0.11462663859128952, + "kept_token_count": 654.6375, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853628039360046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014637198392301798, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.41212158203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0299957275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6300, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.1996876820921898, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.11375908628106117, + "grad_norm": 0.07978985458612442, + "kept_token_count": 654.953125, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853951454162597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014604853000491858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.0580368041992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.72511596679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6310, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1990968883037567, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.12070709839463234, + "grad_norm": 0.06561213731765747, + "kept_token_count": 655.3671875, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844558298587799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015544155426323415, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.3569839477539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8166198730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6320, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19970397800207138, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.130207858979702, + "grad_norm": 0.08659296482801437, + "kept_token_count": 654.68125, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.983798760175705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0162012261338532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.3407836914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.31952209472655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6330, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2007199138402939, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.12752670273184777, + "grad_norm": 0.06972894817590714, + "kept_token_count": 654.240625, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9849064111709595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01509358836337924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.73817749023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.42445678710936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6340, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.20061462819576265, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.12594111487269402, + "grad_norm": 0.08108194917440414, + "kept_token_count": 654.06875, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9852523088455201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014747682213783263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.4269226074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47515563964845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.19888006001710892, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.12618713080883026, + "grad_norm": 0.07940229028463364, + "kept_token_count": 655.5359375, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9837690055370331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016230995208024977, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.53636474609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.65712585449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6360, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20094713866710662, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.12043648660182953, + "grad_norm": 0.09185665100812912, + "kept_token_count": 653.7890625, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9853462636470794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014653727132827043, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.9816162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2336364746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6370, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20010042190551758, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.1292905777692795, + "grad_norm": 0.06827979534864426, + "kept_token_count": 654.9171875, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9837420463562012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016257941164076327, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.8254638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.79429321289064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6380, + "valid_token_count": 818.7484375 + }, + { + "actual_token_mask_ratio": 0.2004014328122139, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.12421090826392174, + "grad_norm": 0.07448936998844147, + "kept_token_count": 654.259375, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.98454709649086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015452893730252982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.1920425415039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0918334960937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6390, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20024965852499008, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.12463559210300446, + "grad_norm": 0.07317709177732468, + "kept_token_count": 654.3421875, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.984116941690445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015883061848580838, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.1405517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5557556152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6400, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19975594878196717, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.1317240133881569, + "grad_norm": 0.09416721761226654, + "kept_token_count": 654.8109375, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850293219089508, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014970682468265295, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.3203750610352, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4664367675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.200656658411026, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.11710207760334015, + "grad_norm": 0.06764529645442963, + "kept_token_count": 653.9375, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.984772789478302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015227200463414192, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.06138305664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5683898925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6420, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.2001877546310425, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.13091323748230935, + "grad_norm": 0.056066494435071945, + "kept_token_count": 654.39375, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847514271736145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015248588100075722, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.7156555175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5607543945313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6430, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20065837800502778, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.12248480916023255, + "grad_norm": 0.06815505772829056, + "kept_token_count": 654.009375, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9843969523906708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015603043232113123, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.8594482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0225006103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6440, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19975325614213943, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.1309954896569252, + "grad_norm": 0.0805315151810646, + "kept_token_count": 654.728125, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9851307213306427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014869286399334668, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.8696319580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5661956787109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20093288719654084, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.13499821051955224, + "grad_norm": 0.07813041657209396, + "kept_token_count": 653.809375, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853862345218658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014613768644630909, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.55090179443357, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5714508056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6460, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20035227090120317, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.12165734767913819, + "grad_norm": 0.09135901927947998, + "kept_token_count": 654.4578125, + "learning_rate": 2.780555866915552e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850224196910858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014977574441581964, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.77129821777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.16759033203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6470, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.1998619928956032, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.12188310846686363, + "grad_norm": 0.08839377760887146, + "kept_token_count": 654.8859375, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9852513492107391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014748660568147897, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.5569412231445, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6324493408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6480, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19927008152008058, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.12105299234390259, + "grad_norm": 0.08657348155975342, + "kept_token_count": 655.2390625, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9851624608039856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01483754375949502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.6818084716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7401519775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6490, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.1991466537117958, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.13247689381241798, + "grad_norm": 0.07077983766794205, + "kept_token_count": 655.14375, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9848732590675354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015126737486571073, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.5736083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6850830078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6500, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.19993941932916642, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.13027724176645278, + "grad_norm": 0.06838902086019516, + "kept_token_count": 654.659375, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845689296722412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015431067440658808, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.5254852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5521484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6510, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.1990855246782303, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.12873316779732705, + "grad_norm": 0.0771130844950676, + "kept_token_count": 655.390625, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9842232406139374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0157767447642982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.36549987792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1038116455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6520, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20028380900621415, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.12652604430913925, + "grad_norm": 0.09692598879337311, + "kept_token_count": 654.0359375, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844968020915985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015503191947937011, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.32770385742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3964050292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6530, + "valid_token_count": 817.8359375 + }, + { + "actual_token_mask_ratio": 0.1999773219227791, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.1228476919233799, + "grad_norm": 0.09971205145120621, + "kept_token_count": 654.625, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9851437985897065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014856199827045202, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.47205047607423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05603942871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6540, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19952362626791, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.11552444845438004, + "grad_norm": 0.08573587238788605, + "kept_token_count": 654.6859375, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9846807003021241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015319292433559895, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.8899841308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.417431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6550, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.20073444843292237, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.1324137605726719, + "grad_norm": 0.0927501916885376, + "kept_token_count": 654.115625, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.984777444601059, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015222562942653894, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.0287719726563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.36460266113284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6560, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.1991366684436798, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.12584415227174758, + "grad_norm": 0.06738486140966415, + "kept_token_count": 655.3765625, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844680726528168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015531925298273564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.5209442138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.16880798339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6570, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20005314499139787, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.10713161379098893, + "grad_norm": 0.0795033797621727, + "kept_token_count": 654.5765625, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.983314722776413, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016685277968645097, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.5772766113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6438049316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6580, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19908479303121568, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.13031615018844606, + "grad_norm": 0.07365508377552032, + "kept_token_count": 655.0890625, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.985117745399475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014882260747253895, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.73853759765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5317779541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6590, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.20010521560907363, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.12436850816011429, + "grad_norm": 0.084774449467659, + "kept_token_count": 654.621875, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9839553117752076, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01604467323049903, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.6411987304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11024475097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6600, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2000002458691597, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.12019828781485557, + "grad_norm": 0.056450702250003815, + "kept_token_count": 654.5078125, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853752255439758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014624784234911203, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.4130493164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.57420959472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.19930929243564605, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.12405341491103172, + "grad_norm": 0.09909038990736008, + "kept_token_count": 655.1375, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9839320540428161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016067935060709714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.0973342895508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8267059326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6620, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2000747427344322, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.1194780245423317, + "grad_norm": 0.06314068287611008, + "kept_token_count": 654.5078125, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9855717658996582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014428231958299875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.6414398193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6827789306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6630, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.19938845336437225, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.12758923023939134, + "grad_norm": 0.07342352718114853, + "kept_token_count": 655.4109375, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847554445266724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01524454653263092, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.9275604248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.64918212890626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6640, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20128016620874406, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.13092688247561454, + "grad_norm": 0.09230849146842957, + "kept_token_count": 653.584375, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844898462295533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01551015703007579, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.50649108886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.33250122070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6650, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20118581056594848, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.11620047017931938, + "grad_norm": 0.10203030705451965, + "kept_token_count": 653.59375, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9841579854488373, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01584202079102397, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.6097045898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3292205810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6660, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.1997241720557213, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.1331183210015297, + "grad_norm": 0.0962526798248291, + "kept_token_count": 654.8078125, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9843746364116669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01562537904828787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.5818298339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.93580627441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6670, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20077164471149445, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.12870488092303276, + "grad_norm": 0.09013215452432632, + "kept_token_count": 653.89375, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847190737724304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015280924551188946, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.217431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3724700927734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6680, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20006286948919297, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.12183277830481529, + "grad_norm": 0.08181530237197876, + "kept_token_count": 654.43125, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9857283234596252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014271666575223208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.99842224121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9886138916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6690, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.19910947233438492, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.12518202960491182, + "grad_norm": 0.06987696141004562, + "kept_token_count": 655.453125, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845387220382691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015461280848830939, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.7821350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.937548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6700, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19990622252225876, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.12601650431752204, + "grad_norm": 0.06310220062732697, + "kept_token_count": 654.8515625, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847908437252044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015209152270108462, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.60611724853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5264068603516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6710, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.19963691681623458, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.14366141632199286, + "grad_norm": 0.058332670480012894, + "kept_token_count": 655.203125, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9848573982715607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01514261420816183, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.5879425048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.94493408203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.19903650432825087, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.13073357418179513, + "grad_norm": 0.06899368017911911, + "kept_token_count": 655.4265625, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847847580909729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015215249452739953, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.02549438476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.27073364257814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6730, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1996736466884613, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.12030378580093384, + "grad_norm": 0.06968177855014801, + "kept_token_count": 654.640625, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9852392256259919, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014760777726769448, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.8205139160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4238647460937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6740, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19994100034236909, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.12833159938454627, + "grad_norm": 0.1291070282459259, + "kept_token_count": 654.8359375, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848103046417236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015189693868160247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.7771331787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.10850830078124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6750, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.20057722479104995, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.12213459089398385, + "grad_norm": 0.08482198417186737, + "kept_token_count": 654.109375, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9848940193653106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01510599572211504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.4762817382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0897918701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6760, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.1996132880449295, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.12371925190091133, + "grad_norm": 0.06339240819215775, + "kept_token_count": 654.9234375, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847402334213257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015259765181690454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.8261199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1468811035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6770, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19909663796424865, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.12566438019275666, + "grad_norm": 0.08716757595539093, + "kept_token_count": 655.1734375, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9854411959648133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014558804500848056, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.3292633056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4081787109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6780, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20049156099557877, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.11989473178982735, + "grad_norm": 0.07649834454059601, + "kept_token_count": 654.2234375, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9853003144264221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014699684083461761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.9462921142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0208251953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6790, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.19934550523757935, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.12369892373681068, + "grad_norm": 0.07885032892227173, + "kept_token_count": 655.065625, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845641314983368, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01543587250635028, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.1411071777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.40830993652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6800, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.1999852940440178, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.13015365228056908, + "grad_norm": 0.07569611072540283, + "kept_token_count": 654.8953125, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845907151699066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015409277100116015, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.80314483642576, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.25267639160154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6810, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.19972104132175444, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.13016935288906098, + "grad_norm": 0.08943735063076019, + "kept_token_count": 654.7109375, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9833000540733338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.016699943970888854, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.08155517578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.38014221191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6820, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.2009444147348404, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.11939423829317093, + "grad_norm": 0.06716472655534744, + "kept_token_count": 653.640625, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856302201747894, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014369780942797662, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.5288360595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7991973876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6830, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.199654783308506, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.12569889947772026, + "grad_norm": 0.10482987016439438, + "kept_token_count": 654.984375, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9838237881660461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01617621900513768, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.48833618164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25184020996096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6840, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2002624213695526, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.12589555904269217, + "grad_norm": 0.08089925348758698, + "kept_token_count": 654.3, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848321080207825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015167888533324004, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.30848999023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.901513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6850, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.20023082047700883, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.1255683831870556, + "grad_norm": 0.061312250792980194, + "kept_token_count": 654.48125, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9844819962978363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015518002212047577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.47191467285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1204345703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6860, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19960381239652633, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.12059333175420761, + "grad_norm": 0.07458136975765228, + "kept_token_count": 654.9140625, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854651272296906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014534866996109485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.7927871704102, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4415344238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6870, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20070713609457017, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.11803323552012443, + "grad_norm": 0.08592721819877625, + "kept_token_count": 654.040625, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858466565608979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014153335057199002, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.92181701660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5129638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6880, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20030311197042466, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.12129874750971795, + "grad_norm": 0.06942304968833923, + "kept_token_count": 654.2109375, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9851595818996429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014840414095669986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.44435729980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1865600585937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6890, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.20095371454954147, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.12362374588847161, + "grad_norm": 0.06261222809553146, + "kept_token_count": 653.9453125, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854601085186004, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014539862796664239, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.49273376464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.62513732910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6900, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20010713785886763, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.13016256168484688, + "grad_norm": 0.07316549122333527, + "kept_token_count": 654.4125, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9841456055641175, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015854399651288986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.58881225585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8113189697266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6910, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.1996345192193985, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.11945354864001274, + "grad_norm": 0.08056721836328506, + "kept_token_count": 654.790625, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.985583633184433, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01441638320684433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.1832763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2847839355469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6920, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.19972358494997025, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.12272592410445213, + "grad_norm": 0.07701923698186874, + "kept_token_count": 654.9671875, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.984619963169098, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015380046702921391, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.7446044921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9389282226563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6930, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.19957620203495025, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.12402117326855659, + "grad_norm": 0.069537453353405, + "kept_token_count": 655.1140625, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9847946763038635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015205312613397836, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.0832122802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5552917480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6940, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.20079289078712464, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.12579960152506828, + "grad_norm": 0.07359917461872101, + "kept_token_count": 654.090625, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9859141349792481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014085859153419733, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.5550140380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1365264892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6950, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20003683418035506, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.1276920199394226, + "grad_norm": 0.07435910403728485, + "kept_token_count": 654.559375, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.984799998998642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015200016926974058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.61619262695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1101348876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6960, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.19949450343847275, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.12265318110585213, + "grad_norm": 0.04974033683538437, + "kept_token_count": 655.084375, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9851697981357574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014830212946981192, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.2993499755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7941009521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19977958053350447, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.12463726475834846, + "grad_norm": 0.0679975375533104, + "kept_token_count": 655.0078125, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9860621571540833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013937848433852196, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.0270721435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.95965576171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6980, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.19933074712753296, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.12064804583787918, + "grad_norm": 0.06612730026245117, + "kept_token_count": 655.1078125, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854660570621491, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014533946849405765, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.9059539794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1145721435547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6990, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19933941960334778, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.1230399914085865, + "grad_norm": 0.07378581911325455, + "kept_token_count": 655.3375, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845091938972473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01549081103876233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.1144104003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40150756835936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7000, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20020731091499328, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.1248562790453434, + "grad_norm": 0.07080260664224625, + "kept_token_count": 654.3171875, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9851570963859558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014842909015715122, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.73142395019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7886260986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7010, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20025103390216828, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.11976161748170852, + "grad_norm": 0.05951573699712753, + "kept_token_count": 654.4859375, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853893041610717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014610692113637924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.38707580566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.15721130371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7020, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19902790188789368, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.11628083810210228, + "grad_norm": 0.07734709978103638, + "kept_token_count": 655.4546875, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850050926208496, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014994913805276155, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.15757446289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4506103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7030, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2002723067998886, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.12041115611791611, + "grad_norm": 0.07436840236186981, + "kept_token_count": 654.315625, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9854027152061462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014597293920814991, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.1421691894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7606689453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7040, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20017356872558595, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.12356726825237274, + "grad_norm": 0.07406224310398102, + "kept_token_count": 654.2515625, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853821039199829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014617882389575242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.5236450195313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0720153808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7050, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.2002467021346092, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.11985979974269867, + "grad_norm": 0.07693439722061157, + "kept_token_count": 654.3109375, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859659790992736, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01403403328731656, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.8972198486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.00229797363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7060, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19935185462236404, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.12036299332976341, + "grad_norm": 0.07114960253238678, + "kept_token_count": 655.2578125, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9842206418514252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015779356379061936, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.3588806152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0823913574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7070, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20027561932802201, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.11664854139089584, + "grad_norm": 0.06205030903220177, + "kept_token_count": 654.171875, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9857691407203675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01423085993155837, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.9442443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1561645507812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20059676617383956, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.11771141961216927, + "grad_norm": 0.0841560885310173, + "kept_token_count": 654.26875, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9857053995132447, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014294589683413506, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.56337280273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.61914978027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7090, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.20000573992729187, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.12170307487249374, + "grad_norm": 0.05504102259874344, + "kept_token_count": 654.521875, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845538318157196, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015446156077086926, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.76104431152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0343414306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7100, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20034362524747848, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.11905507519841194, + "grad_norm": 0.07105517387390137, + "kept_token_count": 654.178125, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861059486865997, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013894054852426053, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.80889282226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9729370117187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7110, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.1989324524998665, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.12488687410950661, + "grad_norm": 0.0838826447725296, + "kept_token_count": 655.3796875, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848437130451202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015156297478824854, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.5622192382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5317687988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7120, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19977832287549974, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.13545494750142098, + "grad_norm": 0.07419145107269287, + "kept_token_count": 654.675, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854691743850708, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014530822075903416, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.7134582519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.99437561035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7130, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19897482842206954, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.1229577675461769, + "grad_norm": 0.07231441140174866, + "kept_token_count": 655.36875, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.985540920495987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014459087140858173, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.9265899658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.65594482421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.19956281036138535, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.118758774548769, + "grad_norm": 0.05634433031082153, + "kept_token_count": 654.915625, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856255233287812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0143744888715446, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.6716278076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41612854003904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7150, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20051816403865813, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.1242963805794716, + "grad_norm": 0.06562062352895737, + "kept_token_count": 654.246875, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861166179180145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013883384689688683, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.8824859619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4204528808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7160, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19962387382984162, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.12540915012359619, + "grad_norm": 0.08242419362068176, + "kept_token_count": 655.1421875, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9862744450569153, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013725543022155761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.5091613769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9631042480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7170, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.20016633570194245, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.1220778375864029, + "grad_norm": 0.0781969428062439, + "kept_token_count": 654.5359375, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856507241725921, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014349280577152967, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.4739990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.68739929199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7180, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20133041590452194, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.13483157530426979, + "grad_norm": 0.07878687977790833, + "kept_token_count": 653.8421875, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9845975041389465, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015402498003095389, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.5525756835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4737182617188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7190, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.20038730949163436, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.1227354273200035, + "grad_norm": 0.06750313937664032, + "kept_token_count": 654.346875, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856870174407959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014312986750155687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.2656951904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7524444580078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7200, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.19992030411958694, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.12198911756277084, + "grad_norm": 0.09733667224645615, + "kept_token_count": 654.5953125, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9853746771812439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014625311642885209, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.61279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5153381347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7210, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.19981672167778014, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.111549411714077, + "grad_norm": 0.07697342336177826, + "kept_token_count": 654.9578125, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854657649993896, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01453424096107483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.17259216308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1031188964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7220, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.19963234663009644, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.13253045827150345, + "grad_norm": 0.07252541184425354, + "kept_token_count": 654.7921875, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9854262053966523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01457381583750248, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.9610107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9432891845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7230, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.19985150694847106, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.13017428517341614, + "grad_norm": 0.06860825419425964, + "kept_token_count": 654.8125, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9865414381027222, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013458560872823, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.30646362304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2214630126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7240, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1987058624625206, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.12536367848515512, + "grad_norm": 0.08131527155637741, + "kept_token_count": 655.809375, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9862696945667266, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01373029425740242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.5389678955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11329650878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7250, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20012072175741197, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.1257132150232792, + "grad_norm": 0.07787443697452545, + "kept_token_count": 654.503125, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859881699085236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014011816401034594, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.2646179199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.98310546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7260, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.20105402320623397, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.1345526844263077, + "grad_norm": 0.07335953414440155, + "kept_token_count": 653.609375, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9853336095809937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01466639144346118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.9809906005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.68516540527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7270, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.1998726472258568, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.1255046509206295, + "grad_norm": 0.05658866837620735, + "kept_token_count": 654.803125, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850106954574585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014989320281893015, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.9862854003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3549377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7280, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20027375519275664, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.1336592599749565, + "grad_norm": 0.07715846598148346, + "kept_token_count": 654.2875, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9843056201934814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015694376733154058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.2551239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9404327392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7290, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.199426107108593, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.11592998802661895, + "grad_norm": 0.05702466517686844, + "kept_token_count": 655.040625, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866528034210205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013347198069095612, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.29554138183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0644561767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7300, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19892801344394684, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.12618364617228509, + "grad_norm": 0.07104117423295975, + "kept_token_count": 655.65625, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9853240966796875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014675916265696287, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.0644561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09754638671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7310, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.2005263552069664, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.12574864625930787, + "grad_norm": 0.06497769802808762, + "kept_token_count": 654.0890625, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9862938165664673, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013706189393997193, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.0915069580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9518768310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7320, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2005741536617279, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.1304336167871952, + "grad_norm": 0.06830234080553055, + "kept_token_count": 653.8421875, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854691684246063, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014530837070196867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.2047485351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.44829711914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7330, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.20028910785913467, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.12614519968628884, + "grad_norm": 0.06813538819551468, + "kept_token_count": 654.540625, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850012004375458, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014998795185238122, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.05812683105466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1313446044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7340, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.20038988143205644, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.13032544478774072, + "grad_norm": 0.07847720384597778, + "kept_token_count": 654.4296875, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9855281591415406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014471846725791692, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.78408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.80625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19960448443889617, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.1287691406905651, + "grad_norm": 0.11934373527765274, + "kept_token_count": 654.903125, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.984403645992279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01559635354205966, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.5312561035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.09800415039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7360, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19939404726028442, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.12993695735931396, + "grad_norm": 0.06263276934623718, + "kept_token_count": 655.1796875, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.985328209400177, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014671786967664958, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.2382873535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.17120971679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7370, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20040091574192048, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.12088469862937927, + "grad_norm": 0.0752253383398056, + "kept_token_count": 654.4078125, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9848114371299743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01518855644389987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.90134887695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7380, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.19905071705579758, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.13286109939217566, + "grad_norm": 0.060727205127477646, + "kept_token_count": 655.5640625, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861313998699188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013868609257042409, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.7361206054687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4744415283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.19919221997261047, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.11188006699085236, + "grad_norm": 0.0713762566447258, + "kept_token_count": 655.3390625, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9855910778045655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014408918190747499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.5135803222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.56687316894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7400, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.20035927295684813, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.12033003866672516, + "grad_norm": 0.06913992017507553, + "kept_token_count": 654.20625, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9871737420558929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012826266791671514, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.3284881591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7920257568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7410, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20025488436222078, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.12154334858059883, + "grad_norm": 0.07483017444610596, + "kept_token_count": 654.4578125, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9865149021148681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01348508670926094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.97791442871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.995947265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7420, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20047388076782227, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.11815062090754509, + "grad_norm": 0.06850708276033401, + "kept_token_count": 654.390625, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856703221797943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014329671300947667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.3335327148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2146697998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7430, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.19897904843091965, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.1155092902481556, + "grad_norm": 0.07315056771039963, + "kept_token_count": 655.3984375, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859881222248077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014011874701827765, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.595263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0790649414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7440, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19911799430847169, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.11577294319868088, + "grad_norm": 0.06509336084127426, + "kept_token_count": 655.4015625, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9847203254699707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015279672108590603, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.3917663574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3171752929687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7450, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20019762665033342, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.13177260980010033, + "grad_norm": 0.08389172703027725, + "kept_token_count": 654.18125, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9862453579902649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013754639029502868, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.17515563964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.50830993652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7460, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.1998303234577179, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.11709078550338745, + "grad_norm": 0.05774926766753197, + "kept_token_count": 654.5765625, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867948710918426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013205130491405726, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.1711456298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.15428466796874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7470, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.200849649310112, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.11492386534810066, + "grad_norm": 0.07814806699752808, + "kept_token_count": 654.1546875, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859766125679016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014023397956043482, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.8029418945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.52422485351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7480, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.20013169944286346, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.11670469790697098, + "grad_norm": 0.06302937865257263, + "kept_token_count": 654.490625, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.985511976480484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01448802724480629, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.5986358642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.85076599121095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7490, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19994289577007293, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.13123959302902222, + "grad_norm": 0.0650111585855484, + "kept_token_count": 654.665625, + "learning_rate": 2.68381231512099e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850085318088532, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01499146968126297, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.81377563476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.77797546386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7500, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20120419710874557, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.11957971379160881, + "grad_norm": 0.08982924371957779, + "kept_token_count": 653.61875, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.015, + "mgd_cos_sim": 0.984981918334961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015018082223832607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.10439147949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3289794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7510, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2003105625510216, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.1216769464313984, + "grad_norm": 0.09830455482006073, + "kept_token_count": 654.328125, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845363914966583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015463615208864212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.2619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2901245117188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7520, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20126895755529403, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.13150618970394135, + "grad_norm": 0.06187848746776581, + "kept_token_count": 653.378125, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9858711779117584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014128815848380327, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.06333923339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.98389587402346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.20113133192062377, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.1192672498524189, + "grad_norm": 0.11938856542110443, + "kept_token_count": 653.9359375, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856102108955384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014389779325574637, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.09031982421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.93006896972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7540, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.19981821328401567, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.12774064540863037, + "grad_norm": 0.07585517317056656, + "kept_token_count": 655.125, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9855323195457458, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014467696286737919, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.98638305664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.82569580078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7550, + "valid_token_count": 818.71875 + }, + { + "actual_token_mask_ratio": 0.19998544603586196, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.13005822896957397, + "grad_norm": 0.08450881391763687, + "kept_token_count": 654.5921875, + "learning_rate": 2.677693837583597e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859719097614288, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014028096478432417, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.92001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.46839904785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7560, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.20030878931283952, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.12486221119761468, + "grad_norm": 0.0653771460056305, + "kept_token_count": 654.50625, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860382616519928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013961746543645858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.5642486572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6705780029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7570, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.20005362033843993, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.11991878971457481, + "grad_norm": 0.10372231155633926, + "kept_token_count": 654.1625, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9852055609226227, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014794444292783737, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.96588745117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40550231933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7580, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.20023391544818878, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.11691751331090927, + "grad_norm": 0.07157283276319504, + "kept_token_count": 654.5328125, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866510689258575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01334893200546503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.5669006347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5417755126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7590, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.20013117045164108, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.1233927220106125, + "grad_norm": 0.0704551488161087, + "kept_token_count": 654.496875, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.986070054769516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013929931353777647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.7300109863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1260986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7600, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20039563477039338, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.12370776236057282, + "grad_norm": 0.0744447186589241, + "kept_token_count": 654.359375, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9852063894271851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014793602004647254, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.3906219482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7836456298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7610, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.19950302690267563, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.12470996379852295, + "grad_norm": 0.07565264403820038, + "kept_token_count": 655.0953125, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866847276687623, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013315270096063614, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.93011474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9515716552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7620, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20026769936084748, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.12466781735420226, + "grad_norm": 0.06195388361811638, + "kept_token_count": 654.4671875, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9860959649085999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013904028292745351, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.6047821044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4246887207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7630, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.1993558555841446, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.12276363298296929, + "grad_norm": 0.0626085177063942, + "kept_token_count": 655.48125, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861657321453094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013834260497242212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.2725372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1978393554688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7640, + "valid_token_count": 818.7 + }, + { + "actual_token_mask_ratio": 0.1995432510972023, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.11381430178880692, + "grad_norm": 0.07950645685195923, + "kept_token_count": 654.834375, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9849557340145111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015044248290359974, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.921484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8134796142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7650, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20075699985027312, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.13452538028359412, + "grad_norm": 0.062158819288015366, + "kept_token_count": 654.103125, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856443643569947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014355618320405484, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.8695465087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2265655517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7660, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19998958855867385, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.12981499284505843, + "grad_norm": 0.12680847942829132, + "kept_token_count": 654.74375, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858473122119904, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014152693096548319, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.9265533447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11907958984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7670, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.1991179794073105, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.12492617070674897, + "grad_norm": 0.062981978058815, + "kept_token_count": 655.41875, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9868488252162934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013151164073497057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.9501525878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7953186035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7680, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19972898066043854, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.12362183332443237, + "grad_norm": 0.06565233319997787, + "kept_token_count": 654.828125, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.986091923713684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013908062130212784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.62864074707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0702392578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20078511387109757, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.13115825057029723, + "grad_norm": 0.06799348443746567, + "kept_token_count": 653.8484375, + "learning_rate": 2.663217516246862e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860098481178283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013990151509642602, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.55237731933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2419799804687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7700, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2007964640855789, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.12077411189675331, + "grad_norm": 0.0748274102807045, + "kept_token_count": 653.815625, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9857433259487152, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014256674982607364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.1407531738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7148864746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7710, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19973184615373613, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.11613232344388962, + "grad_norm": 0.08341927826404572, + "kept_token_count": 654.915625, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9857225358486176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01427746070548892, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.87401428222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.22910766601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7720, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19957042038440703, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.12194929197430611, + "grad_norm": 0.08762645721435547, + "kept_token_count": 654.9203125, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861929714679718, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013807031884789467, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.9819000244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4926330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7730, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.20010388940572738, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.12197805196046829, + "grad_norm": 0.07107436656951904, + "kept_token_count": 654.878125, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9862061500549316, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013793842680752278, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.7986694335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.291162109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7740, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.19937609881162643, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.11901117339730263, + "grad_norm": 0.0723741203546524, + "kept_token_count": 655.25625, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866296529769898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01337034059688449, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.81833801269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2731170654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7750, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20066163092851638, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.12429754734039307, + "grad_norm": 0.0583370141685009, + "kept_token_count": 654.0140625, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9849212408065796, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015078757610172033, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.1216064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.682421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7760, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20002058297395706, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.1303536392748356, + "grad_norm": 0.10718797147274017, + "kept_token_count": 654.80625, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9854603350162506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014539662562310696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.11724853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3834259033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7770, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.1997959718108177, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.12229845151305199, + "grad_norm": 0.07910092920064926, + "kept_token_count": 654.9421875, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856886386871337, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014311377145349979, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.6686950683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9047912597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2008868858218193, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.125042824447155, + "grad_norm": 0.055758316069841385, + "kept_token_count": 653.759375, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861294388771057, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013870566803961992, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.49383239746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3411437988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7790, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20001684576272966, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.12756813913583756, + "grad_norm": 0.09261798113584518, + "kept_token_count": 654.4984375, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9863363623619079, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01366363987326622, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.04779663085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1958770751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7800, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.20024160742759706, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.12433127462863922, + "grad_norm": 0.07021290063858032, + "kept_token_count": 654.534375, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9845167398452759, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015483257453888654, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.082861328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.62274475097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7810, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.1997716471552849, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.12033166661858559, + "grad_norm": 0.07754666358232498, + "kept_token_count": 654.9078125, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868913531303406, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013108631875365973, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.246826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0318237304688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7820, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2009078308939934, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.12510673552751542, + "grad_norm": 0.0579821802675724, + "kept_token_count": 653.946875, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858429372310639, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014157064724713564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.1026123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3936065673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7830, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20005747973918914, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.12463891804218293, + "grad_norm": 0.06533083319664001, + "kept_token_count": 654.8640625, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866453468799591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013354649487882852, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.91185607910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7658752441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7840, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.20022148340940477, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.11217721551656723, + "grad_norm": 0.09807897359132767, + "kept_token_count": 654.409375, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856326162815094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014367383252829314, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.3157989501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9519317626953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7850, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20045480281114578, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.11872718781232834, + "grad_norm": 0.06409435719251633, + "kept_token_count": 654.228125, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867392241954803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013260770589113235, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.15150756835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.74644470214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.20024905055761338, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.11804241836071014, + "grad_norm": 0.06971584260463715, + "kept_token_count": 654.2921875, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858197808265686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014180229045450687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.30126037597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8673614501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7870, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19912694543600082, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.12481152936816216, + "grad_norm": 0.08844458311796188, + "kept_token_count": 655.2296875, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.986346161365509, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013653850089758635, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.86590881347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1049743652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7880, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.19953396916389465, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.1296778917312622, + "grad_norm": 0.084893137216568, + "kept_token_count": 654.8328125, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858251750469208, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014174821600317956, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.21177673339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6056182861328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7890, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.20035915970802307, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.1241645984351635, + "grad_norm": 0.06950999796390533, + "kept_token_count": 654.1890625, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860489249229432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013951086159795522, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.8525817871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8161163330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7900, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.20015753656625748, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.11647913604974747, + "grad_norm": 0.062499091029167175, + "kept_token_count": 654.4890625, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.987226814031601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01277318773791194, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.8952270507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.31481018066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7910, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20036335438489913, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.11793896406888962, + "grad_norm": 0.05126022920012474, + "kept_token_count": 654.259375, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865776717662811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013422332424670457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.01604919433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0712432861328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7920, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.19894177168607713, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.1142346553504467, + "grad_norm": 0.06744623929262161, + "kept_token_count": 655.5484375, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867350041866303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013265005126595498, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.22247619628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.29814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7930, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19886847585439682, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.12420498877763748, + "grad_norm": 0.08849043399095535, + "kept_token_count": 655.340625, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9863111197948455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01368887210264802, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.3802032470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0883056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7940, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2012089729309082, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.12372941374778748, + "grad_norm": 0.07859627902507782, + "kept_token_count": 653.6, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9855051279067993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014494871348142624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.71156005859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.689111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.200236177444458, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.11886081248521804, + "grad_norm": 0.06829285621643066, + "kept_token_count": 654.346875, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9862839043140411, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013716079015284776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.0509918212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.42406005859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7960, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2006101056933403, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.12990358769893645, + "grad_norm": 0.07119517773389816, + "kept_token_count": 654.06875, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856845378875733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014315471518784761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.1462097167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.75836181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19962187260389327, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.12128631472587585, + "grad_norm": 0.11737477034330368, + "kept_token_count": 654.775, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865827620029449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01341723995283246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.3636901855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3302307128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7980, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.1997327461838722, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.12411883547902107, + "grad_norm": 0.07881695032119751, + "kept_token_count": 654.865625, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9843276977539063, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015672282595187426, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.3712554931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0443572998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7990, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.19948261231184006, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.11947977170348167, + "grad_norm": 0.09872271120548248, + "kept_token_count": 655.0390625, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9848844707012177, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.015115530882030725, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.97124938964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83250732421874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8000, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19995402097702025, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.11813111677765846, + "grad_norm": 0.0730757936835289, + "kept_token_count": 654.6796875, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865574777126312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013442519586533308, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.8458953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.34111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8010, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.2003251865506172, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.11869483664631844, + "grad_norm": 0.08125797659158707, + "kept_token_count": 654.428125, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865531086921692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013446885254234075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.93251342773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.54982299804686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8020, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.200611674785614, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.12219352498650551, + "grad_norm": 0.06224668398499489, + "kept_token_count": 654.2234375, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858179271221161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014182075299322605, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.7786437988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.570361328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8030, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.19950360506772996, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.12128567844629287, + "grad_norm": 0.08089417964220047, + "kept_token_count": 655.3359375, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861474514007569, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013852558005601168, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.98651733398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3135986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8040, + "valid_token_count": 818.665625 + }, + { + "actual_token_mask_ratio": 0.20012476593255996, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.12130651324987411, + "grad_norm": 0.07906796783208847, + "kept_token_count": 654.603125, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9852416276931762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014758371189236641, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.9207397460938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8954162597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8050, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2008661985397339, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.11806179210543633, + "grad_norm": 0.05438096821308136, + "kept_token_count": 653.953125, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866994798183442, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013300536014139653, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.9047515869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7585418701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8060, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20037904232740403, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.11908326894044877, + "grad_norm": 0.05493491515517235, + "kept_token_count": 654.165625, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9862523972988129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013747600093483924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.5342193603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.56728515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8070, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19899928271770478, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.12676916792988777, + "grad_norm": 0.06230424344539642, + "kept_token_count": 655.521875, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866178035736084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013382216449826956, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.81776123046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.50147399902346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8080, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.20090369284152984, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.12544819638133048, + "grad_norm": 0.06980791687965393, + "kept_token_count": 653.8828125, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864113450050354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013588646426796913, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.72796325683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3417999267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8090, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.19981090873479843, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.12596465572714804, + "grad_norm": 0.06946580857038498, + "kept_token_count": 654.6609375, + "learning_rate": 2.620342168475869e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9850448846817017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01495511932298541, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.32233276367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7288818359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8100, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19960069209337233, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.1268316388130188, + "grad_norm": 0.08320954442024231, + "kept_token_count": 655.0671875, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9862298846244812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013770126178860665, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.33771057128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9427856445312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8110, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.20037296861410142, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.11895509958267211, + "grad_norm": 0.07946350425481796, + "kept_token_count": 653.990625, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9849267125129699, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01507329111918807, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.9689880371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.86854858398436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8120, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.19987538903951646, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.13059312403202056, + "grad_norm": 0.09376577287912369, + "kept_token_count": 654.5953125, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861642062664032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013835806865245104, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.6967834472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32777099609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8130, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19941334575414657, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.126967391371727, + "grad_norm": 0.06030500680208206, + "kept_token_count": 654.9828125, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858010947704315, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014198900107294321, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.6177978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89268798828124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8140, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19958522617816926, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.12856974378228186, + "grad_norm": 0.09249140322208405, + "kept_token_count": 654.921875, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866904258728028, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013309573754668236, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.6389892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5988311767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8150, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20010476559400558, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.11855574771761894, + "grad_norm": 0.08214418590068817, + "kept_token_count": 654.6359375, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858071148395539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01419287510216236, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.89613647460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9529205322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8160, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20068154931068422, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.12347392290830612, + "grad_norm": 0.09024564921855927, + "kept_token_count": 654.3078125, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9859189808368682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014081014785915613, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.1743896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.08826599121096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8170, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.20054319649934768, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.11422340646386146, + "grad_norm": 0.06593595445156097, + "kept_token_count": 653.8421875, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869374573230744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01306253531947732, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.0033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.34910583496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8180, + "valid_token_count": 817.8578125 + }, + { + "actual_token_mask_ratio": 0.20043392926454545, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.13869686648249627, + "grad_norm": 0.08218709379434586, + "kept_token_count": 654.2703125, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9850763797760009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01492360169067979, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.35138244628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4333770751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8190, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.1997198447585106, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.11708761677145958, + "grad_norm": 0.06544848531484604, + "kept_token_count": 654.7796875, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9865413665771484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01345863165333867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.1221099853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6593841552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.1994474396109581, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.11798896118998528, + "grad_norm": 0.07288827747106552, + "kept_token_count": 654.934375, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858041822910308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014195819944143295, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.73052368164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10059814453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8210, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.19950605034828187, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.12178626134991646, + "grad_norm": 0.0792599767446518, + "kept_token_count": 655.078125, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865744888782502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013425524067133664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.6814331054687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.21152648925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2006279155611992, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.12382354959845543, + "grad_norm": 0.0726148709654808, + "kept_token_count": 654.2703125, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9863055527210236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013694435451179742, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.8344421386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3586181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8230, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19970194548368453, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.1239679366350174, + "grad_norm": 0.06911586970090866, + "kept_token_count": 654.9546875, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9860567450523376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013943244144320488, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.52513732910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.63352966308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8240, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.19997987896203995, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.12152046263217926, + "grad_norm": 0.07309798896312714, + "kept_token_count": 654.7203125, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865968823432922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013403123244643211, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.1282196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7679504394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8250, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.19951860308647157, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.12276670187711716, + "grad_norm": 0.07321343570947647, + "kept_token_count": 654.64375, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9858520925045013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014147902373224497, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.4523010253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4356994628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8260, + "valid_token_count": 817.8125 + }, + { + "actual_token_mask_ratio": 0.20026632994413376, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.13006725013256074, + "grad_norm": 0.06096573919057846, + "kept_token_count": 654.5109375, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871075749397278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012892431672662496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.7713653564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01795959472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8270, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19991546720266343, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.12167288810014724, + "grad_norm": 0.0746486708521843, + "kept_token_count": 654.8125, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867168843746186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013283118605613708, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.4427917480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.11866760253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8280, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.19918460249900818, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.12276647612452507, + "grad_norm": 0.07223143428564072, + "kept_token_count": 655.5953125, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9857739984989167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014225996006280184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.19261474609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4803741455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8290, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.19930745512247086, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.12617707028985023, + "grad_norm": 0.09335838258266449, + "kept_token_count": 655.2109375, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9857788860797883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01422111364081502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.3794311523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.85630493164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8300, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.1998535543680191, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.12582357823848725, + "grad_norm": 0.07701554894447327, + "kept_token_count": 654.825, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9868203341960907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013179656583815813, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.2530548095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8512023925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8310, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.1991359442472458, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.12283123433589935, + "grad_norm": 0.06469330191612244, + "kept_token_count": 655.2625, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9854061961174011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014593797363340854, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.2581451416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59388427734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8320, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.19944994747638703, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.12236317917704582, + "grad_norm": 0.06306019425392151, + "kept_token_count": 654.8, + "learning_rate": 2.594697202771014e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869662463665009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013033743016421795, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.4282287597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0754150390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8330, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.20000219196081162, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.12308142557740212, + "grad_norm": 0.07405359297990799, + "kept_token_count": 654.509375, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861502945423126, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013849691953510046, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.40748291015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.438037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8340, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.20031391829252243, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.1211918018758297, + "grad_norm": 0.09683039039373398, + "kept_token_count": 654.2109375, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9856217265129089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014378286432474852, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.10223083496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6939392089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8350, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19926720559597016, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.12614108771085739, + "grad_norm": 0.056139785796403885, + "kept_token_count": 655.24375, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864412784576416, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013558732531964778, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.4615020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3023712158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8360, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20094924420118332, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.11465019062161445, + "grad_norm": 0.05187589302659035, + "kept_token_count": 653.5890625, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861521661281586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013847841694951058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.18837585449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.19444274902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8370, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.2005514159798622, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.12037424221634865, + "grad_norm": 0.057185497134923935, + "kept_token_count": 654.14375, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9863650321960449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013634964171797038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.26804809570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5023651123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8380, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20002651810646058, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.1223487950861454, + "grad_norm": 0.057584986090660095, + "kept_token_count": 654.4734375, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869227170944214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01307728560641408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.18886108398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.309326171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8390, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20044111013412474, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.12527921050786972, + "grad_norm": 0.05957061052322388, + "kept_token_count": 654.1796875, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876304745674134, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012369523849338293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.12343139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2465087890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8400, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.2005735903978348, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.11928933709859849, + "grad_norm": 0.0851592943072319, + "kept_token_count": 654.146875, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866403043270111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01335968980565667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.15967712402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.05042419433596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8410, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19916707277297974, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.125258120149374, + "grad_norm": 0.07746371626853943, + "kept_token_count": 655.240625, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864056468009949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013594352081418038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.89182739257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9718048095703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8420, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.1999307245016098, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.126707324385643, + "grad_norm": 0.05015348270535469, + "kept_token_count": 654.8265625, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9863914966583252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01360849617049098, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.6491668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.92699279785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8430, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.20062485486268997, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.12724282070994378, + "grad_norm": 0.06830327957868576, + "kept_token_count": 654.1453125, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869042098522186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01309579135850072, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.7787689208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8490234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8440, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20052549093961716, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.12916378900408745, + "grad_norm": 0.05572160705924034, + "kept_token_count": 654.4078125, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864202439785004, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013579756394028664, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.7692932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.70050048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8450, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19931185692548753, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.1296609453856945, + "grad_norm": 0.059763044118881226, + "kept_token_count": 655.23125, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866783320903778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01332167536020279, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.827880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8625823974609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8460, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19940400421619414, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.13272615373134614, + "grad_norm": 0.07094182819128036, + "kept_token_count": 654.825, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870854616165161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01291453130543232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.9884887695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.94915161132815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8470, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.20096643418073654, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.12818572521209717, + "grad_norm": 0.06536970287561417, + "kept_token_count": 653.8953125, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866315245628356, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013368481118232012, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.7276275634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9783660888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8480, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19924393445253372, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.12624313831329345, + "grad_norm": 0.07885119318962097, + "kept_token_count": 655.0578125, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866258442401886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013374153152108192, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.18195190429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3982330322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8490, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.2005691945552826, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.13278904110193251, + "grad_norm": 0.08166630566120148, + "kept_token_count": 654.00625, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866617739200592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013338232785463333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.85052490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6456359863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8500, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20080700963735582, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.12344300150871276, + "grad_norm": 0.10705889016389847, + "kept_token_count": 653.878125, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868546247482299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013145373202860355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.3045135498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0498382568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8510, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.19991809129714966, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.12894269078969955, + "grad_norm": 0.06906288117170334, + "kept_token_count": 654.759375, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.985850703716278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014149290136992931, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.6726013183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9680572509766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8520, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.1997276559472084, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.13090729489922523, + "grad_norm": 0.056685831397771835, + "kept_token_count": 654.9453125, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870036840438843, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012996311020106077, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.2864593505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.84560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8530, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2005839690566063, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.13764648884534836, + "grad_norm": 0.07205823063850403, + "kept_token_count": 653.896875, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867628455162049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013237149454653263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.2375854492187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.62946166992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8540, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.20008532255887984, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.12148097157478333, + "grad_norm": 0.06110919639468193, + "kept_token_count": 654.6890625, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873170733451844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012682927772402763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.14967346191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1726348876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8550, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20034887492656708, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.13549339547753333, + "grad_norm": 0.07464033365249634, + "kept_token_count": 654.2734375, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9862195909023285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013780410587787627, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.84795532226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9977264404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8560, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20019474476575852, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.13238713517785072, + "grad_norm": 0.059782933443784714, + "kept_token_count": 654.5421875, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858435094356537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014156486745923758, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.5661560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.91371765136716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8570, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20009854286909104, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.12392670586705208, + "grad_norm": 0.07659382373094559, + "kept_token_count": 654.7125, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873694598674774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012630545999854803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.042431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.09496154785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8580, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.1996416136622429, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.11742288395762443, + "grad_norm": 0.06608172506093979, + "kept_token_count": 654.9640625, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866273403167725, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013372679334133863, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.6223205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.77333984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2003335699439049, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.12770009115338327, + "grad_norm": 0.07589085400104523, + "kept_token_count": 654.5328125, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864046335220337, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013595371693372726, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.4396697998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9491424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8600, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20007613748311998, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.12815267741680145, + "grad_norm": 0.06769581139087677, + "kept_token_count": 654.821875, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.985612291097641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01438769670203328, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.2673370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9360748291016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8610, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.19981729686260224, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.1321623258292675, + "grad_norm": 0.0527905635535717, + "kept_token_count": 654.76875, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9856687188148499, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014331276528537273, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.05279541015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.96177978515624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8620, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20024906545877458, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.1282563880085945, + "grad_norm": 0.04829992726445198, + "kept_token_count": 654.4265625, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872807800769806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012719220016151667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.8442687988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81788940429686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8630, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20050135105848313, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.13229512125253678, + "grad_norm": 0.06450562179088593, + "kept_token_count": 654.2359375, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866467535495758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013353240769356489, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.7960144042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.13243408203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8640, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20019212514162063, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.13210833668708802, + "grad_norm": 0.059116873890161514, + "kept_token_count": 654.3578125, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9871555149555207, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01284448355436325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.6587188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.48326416015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.1998631998896599, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.12390158101916313, + "grad_norm": 0.09240758419036865, + "kept_token_count": 654.6796875, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870984256267548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012901566457003355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.4427795410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1332611083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8660, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20009436309337617, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.13785873129963874, + "grad_norm": 0.10030995309352875, + "kept_token_count": 654.396875, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9857931137084961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014206877071410417, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.3521301269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.14024047851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8670, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19977087378501893, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.1347399652004242, + "grad_norm": 0.09271569550037384, + "kept_token_count": 654.7328125, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864001035690307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013599893730133771, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.5494659423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.98133850097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8680, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19998684376478196, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.11928443387150764, + "grad_norm": 0.08159396052360535, + "kept_token_count": 654.421875, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870502173900604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012949780840426684, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.8445983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.76988830566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8690, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.1993314355611801, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.11780527457594872, + "grad_norm": 0.07120627909898758, + "kept_token_count": 655.215625, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9864833354949951, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013516660500317811, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.73455505371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7886901855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19985199272632598, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.13101101368665696, + "grad_norm": 0.07116349041461945, + "kept_token_count": 654.6171875, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9855862021446228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014413800928741694, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.8499755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.13475646972654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8710, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19907332956790924, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.1280365578830242, + "grad_norm": 0.06534446775913239, + "kept_token_count": 655.5390625, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.986328637599945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013671379256993531, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.67491455078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7753173828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8720, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.19856025278568268, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.12495498284697533, + "grad_norm": 0.07513933628797531, + "kept_token_count": 656.165625, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872619390487671, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012738059926778077, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.46495666503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.93018493652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8730, + "valid_token_count": 818.7296875 + }, + { + "actual_token_mask_ratio": 0.19955336302518845, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.12881512492895125, + "grad_norm": 0.07326658815145493, + "kept_token_count": 655.003125, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872528374195099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012747163698077203, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.18017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5968505859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8740, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19964338093996048, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.12188283130526542, + "grad_norm": 0.06779710948467255, + "kept_token_count": 654.78125, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858312368392944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014168754126876593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.15185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.41748046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8750, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20087334960699083, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.12052212059497833, + "grad_norm": 0.06549739092588425, + "kept_token_count": 654.115625, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868982195854187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013101790193468332, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.5184814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.79149475097654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8760, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.2003987267613411, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.13372078612446786, + "grad_norm": 0.0767097994685173, + "kept_token_count": 654.1390625, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9871750056743622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012824999913573266, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.2497497558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0028594970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8770, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.19969638288021088, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.12694049775600433, + "grad_norm": 0.0757773295044899, + "kept_token_count": 654.9984375, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9865466535091401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01345335179939866, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.9242218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2150817871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8780, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.19956283122301102, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.12687264680862426, + "grad_norm": 0.06582015007734299, + "kept_token_count": 654.959375, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9864880025386811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013511980697512627, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.65496826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.79530639648436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8790, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.20044746547937392, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.12289382070302964, + "grad_norm": 0.06589556485414505, + "kept_token_count": 654.3140625, + "learning_rate": 2.540122599063442e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860283076763153, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01397169465199113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.5416229248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49346313476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2004301980137825, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.13158441260457038, + "grad_norm": 0.059894394129514694, + "kept_token_count": 654.3671875, + "learning_rate": 2.538930578929859e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9859812617301941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014018737338483333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.63734130859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9108367919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8810, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19993137270212175, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.14219469726085662, + "grad_norm": 0.05416127294301987, + "kept_token_count": 654.7328125, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867528975009918, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013247091509401798, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.0570983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.83013305664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8820, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20012660473585128, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.13270987942814827, + "grad_norm": 0.0924372747540474, + "kept_token_count": 654.6453125, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865766227245331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013423377182334661, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.5483093261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0920654296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19916694462299347, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.13263773769140244, + "grad_norm": 0.058454688638448715, + "kept_token_count": 655.29375, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.987232917547226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012767091114073991, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.8435882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8614776611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8840, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2002586007118225, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.1330266408622265, + "grad_norm": 0.06569920480251312, + "kept_token_count": 654.3640625, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9865075409412384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013492448534816504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.49965515136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83241577148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8850, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20015917718410492, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.1266696996986866, + "grad_norm": 0.0605950690805912, + "kept_token_count": 654.50625, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872997999191284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012700182292610407, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.8328399658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.90296325683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8860, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.1995982602238655, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.1303028218448162, + "grad_norm": 0.09414342790842056, + "kept_token_count": 654.909375, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9861011981964112, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013898799009621144, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.3496887207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.56447143554686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8870, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.1999412953853607, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.11796625480055808, + "grad_norm": 0.06920180469751358, + "kept_token_count": 654.75, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866649270057678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013335058744996787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.76064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4690643310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19961447268724442, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.12377545982599258, + "grad_norm": 0.05579782649874687, + "kept_token_count": 655.003125, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876497387886047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012350270431488753, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.83384704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1131072998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8890, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20023462921380997, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.11715849861502647, + "grad_norm": 0.054711099714040756, + "kept_token_count": 654.4453125, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864481329917908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013551865611225367, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.9804229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0275085449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8900, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19982328563928603, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.1257091611623764, + "grad_norm": 0.06671800464391708, + "kept_token_count": 654.8625, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869288146495819, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013071183487772942, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.4313110351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.20374450683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19946506172418593, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.12963614165782927, + "grad_norm": 0.07630268484354019, + "kept_token_count": 654.8828125, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860409677028656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013959013856947422, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.95391540527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4752227783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8920, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.20044889748096467, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.13557653725147248, + "grad_norm": 0.09102673083543777, + "kept_token_count": 654.203125, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9858512461185456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014148731529712678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.13470153808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7793731689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8930, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2004839614033699, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.13632472157478331, + "grad_norm": 0.07490961253643036, + "kept_token_count": 654.3890625, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866364240646363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01336357370018959, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.6874969482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.55247497558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8940, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.19975619316101073, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.13083385825157165, + "grad_norm": 0.060844432562589645, + "kept_token_count": 655.059375, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871011197566986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012898879125714302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.18243713378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8464630126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8950, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.20087676346302033, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.13314544707536696, + "grad_norm": 0.0816224217414856, + "kept_token_count": 653.7953125, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865709543228149, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013429053034633398, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.4368499755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15528259277346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8960, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.20024032294750213, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.1266482427716255, + "grad_norm": 0.056654542684555054, + "kept_token_count": 654.2984375, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873630106449127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012636979576200247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.2062255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87701416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8970, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20004567205905915, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.12190793603658676, + "grad_norm": 0.05849136412143707, + "kept_token_count": 654.446875, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9864822208881379, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013517768029123545, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.17645263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2523193359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8980, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2005978673696518, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.1224712997674942, + "grad_norm": 0.061265729367733, + "kept_token_count": 654.2015625, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869721055030822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013027901109308005, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.71034851074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9294006347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8990, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20112180411815644, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.1221812754869461, + "grad_norm": 0.054434485733509064, + "kept_token_count": 653.6234375, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873218894004822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012678104266524316, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.389794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.46222229003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9000, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.200207656621933, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.12971914038062096, + "grad_norm": 0.07468314468860626, + "kept_token_count": 654.690625, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867006540298462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013299343455582858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.30401000976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10255432128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9010, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.20011476129293443, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.12153659090399742, + "grad_norm": 0.07774633914232254, + "kept_token_count": 654.4875, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867259860038757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013274027686566114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.2681182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3763916015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9020, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20029129683971406, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.12995758578181266, + "grad_norm": 0.07957150787115097, + "kept_token_count": 654.425, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858250200748444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.014174985140562058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.2339752197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0201873779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9030, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19932012408971786, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.1293223112821579, + "grad_norm": 0.06394323706626892, + "kept_token_count": 655.1375, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860090136528015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01399098252877593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.42071838378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6618316650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9040, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.20036592185497284, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.12040062397718429, + "grad_norm": 0.06110797077417374, + "kept_token_count": 654.615625, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9863334894180298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013666515238583088, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.96083068847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8026550292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9050, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20067256242036818, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.13778457194566726, + "grad_norm": 0.059970345348119736, + "kept_token_count": 654.3046875, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871210098266602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01287898113951087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.5203491210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.93871154785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9060, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.19991825371980668, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.1241820402443409, + "grad_norm": 0.0694858580827713, + "kept_token_count": 654.68125, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9860310137271882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013968984223902225, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.1347381591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0859313964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9070, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20032841414213182, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.12276459038257599, + "grad_norm": 0.06280242651700974, + "kept_token_count": 654.6375, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.987101924419403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012898063473403453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.83191833496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4382629394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9080, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.20080344527959823, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.11989417299628258, + "grad_norm": 0.08360844850540161, + "kept_token_count": 654.1171875, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866857171058655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013314284849911928, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.50341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.73780822753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9090, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.20007078796625138, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.12570910006761551, + "grad_norm": 0.07518161833286285, + "kept_token_count": 654.58125, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871453166007995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012854696530848742, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.14178161621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89923400878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9100, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20008527040481566, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.12244623899459839, + "grad_norm": 0.05931688845157623, + "kept_token_count": 654.59375, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866823732852936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013317620661109687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.6551574707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1021392822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2005843088030815, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.12435592934489251, + "grad_norm": 0.06469199806451797, + "kept_token_count": 654.1265625, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867610692977905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013238922227174044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.1943786621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0980987548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9120, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19982910752296448, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.12588490322232246, + "grad_norm": 0.05352313816547394, + "kept_token_count": 654.875, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.987883573770523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012116430327296257, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.22312316894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1490417480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9130, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.19997774213552474, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.12288216203451156, + "grad_norm": 0.072535939514637, + "kept_token_count": 654.7296875, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865859031677247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013414094690233469, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.6927001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9798645019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9140, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.1988701343536377, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.1330782376229763, + "grad_norm": 0.06485480070114136, + "kept_token_count": 655.6140625, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.98644118309021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013558805454522371, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.91350708007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78953247070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9150, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19949536621570588, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.11880555748939514, + "grad_norm": 0.06271886825561523, + "kept_token_count": 654.9890625, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9863565564155579, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013643444050103426, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.2560607910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5124084472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9160, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.20051679164171218, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.12475157976150512, + "grad_norm": 0.06879591196775436, + "kept_token_count": 654.16875, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.013, + "mgd_cos_sim": 0.987006688117981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012993319611996412, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.6264190673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4242462158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9170, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19977059215307236, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.11885418146848678, + "grad_norm": 0.05504018813371658, + "kept_token_count": 654.571875, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.986719411611557, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01328059183433652, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.5213928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3080780029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9180, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.20085955858230592, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.1314316600561142, + "grad_norm": 0.05938762053847313, + "kept_token_count": 654.0453125, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869398355484009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013060173112899065, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.8624206542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.54359130859376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9190, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19891666769981384, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.12290275767445565, + "grad_norm": 0.07664220780134201, + "kept_token_count": 655.4265625, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869802713394165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013019723165780307, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.2786163330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.34913024902346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20020783841609954, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.12399461716413498, + "grad_norm": 0.09251512587070465, + "kept_token_count": 654.3109375, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869551479816436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013044849969446659, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.75531005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.59617614746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9210, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.20077090114355087, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.13182856142520905, + "grad_norm": 0.06928049027919769, + "kept_token_count": 653.928125, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.986470514535904, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013529479596763849, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.7808013916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1216339111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9220, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20005444437265396, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.12835954874753952, + "grad_norm": 0.06459570676088333, + "kept_token_count": 654.7265625, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9858472406864166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01415275949984789, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.5831237792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6159362792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9230, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.1999094918370247, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.1164953775703907, + "grad_norm": 0.07686100155115128, + "kept_token_count": 654.7546875, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867388963699341, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013261093851178884, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.7485321044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7181701660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9240, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20095693320035934, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.11734220460057258, + "grad_norm": 0.06015140563249588, + "kept_token_count": 653.90625, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873583316802979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012641685642302036, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.8803314208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.35154724121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9250, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2001728668808937, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.12058327272534371, + "grad_norm": 0.058069564402103424, + "kept_token_count": 654.38125, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9874175667762757, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01258243527263403, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.3092956542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84556579589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9260, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19979844689369203, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.12123143821954727, + "grad_norm": 0.057069312781095505, + "kept_token_count": 655.1796875, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876413702964782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0123586380854249, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.3236938476563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1648681640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9270, + "valid_token_count": 818.765625 + }, + { + "actual_token_mask_ratio": 0.2004643201828003, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.12100266590714455, + "grad_norm": 0.06537666916847229, + "kept_token_count": 654.428125, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870587944984436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012941211648285388, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.61265258789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.20339050292966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9280, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2003219962120056, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.1211477518081665, + "grad_norm": 0.05871133878827095, + "kept_token_count": 654.2609375, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881599128246308, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011840114276856184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.1059814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.37421264648435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9290, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.1994229033589363, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.12663343995809556, + "grad_norm": 0.07178305089473724, + "kept_token_count": 655.00625, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875949561595917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012405036576092243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.83424377441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41050415039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9300, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.19982362985610963, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.12708572447299957, + "grad_norm": 0.054633334279060364, + "kept_token_count": 654.4921875, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871460855007171, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012853906862437725, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.4580505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4366424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9310, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.20014430284500123, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.11851639077067375, + "grad_norm": 0.06801316142082214, + "kept_token_count": 654.6859375, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869768619537354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013023128360509872, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.3086669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4272796630859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9320, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.1998566657304764, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.12456714436411857, + "grad_norm": 0.07169593870639801, + "kept_token_count": 654.771875, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866479396820068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013352056220173835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.4351837158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3406707763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9330, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2010179117321968, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.13372819796204566, + "grad_norm": 0.06124584376811981, + "kept_token_count": 653.8390625, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.987090277671814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012909716740250588, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.9517364501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2363647460937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9340, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20083755254745483, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.13063646480441093, + "grad_norm": 0.07033180445432663, + "kept_token_count": 653.978125, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868806421756744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013119356986135245, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6402893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.97276916503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20045932680368422, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.11211714819073677, + "grad_norm": 0.09307729452848434, + "kept_token_count": 654.31875, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872171878814697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01278282068669796, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.0220153808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1951873779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20082101970911026, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.11030897572636604, + "grad_norm": 0.07195943593978882, + "kept_token_count": 653.940625, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868687748908996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0131312295794487, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.115966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2020782470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19938216209411622, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.11522713378071785, + "grad_norm": 0.05761045590043068, + "kept_token_count": 655.090625, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9876908838748932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012309108022600412, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.3068450927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7664337158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20023271739482879, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.11981004402041436, + "grad_norm": 0.08148408681154251, + "kept_token_count": 654.5859375, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.986317491531372, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01368250185623765, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.9574401855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4405822753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9390, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.19963601529598235, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.1236296348273754, + "grad_norm": 0.06694896519184113, + "kept_token_count": 654.9515625, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866755664348602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013324435520917177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.9098815917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70069885253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9400, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.19900622665882112, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.13037203773856162, + "grad_norm": 0.07485076785087585, + "kept_token_count": 655.6203125, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864377915859223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013562214002013207, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.9714019775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.96839904785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9410, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.19950202256441116, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.1221973642706871, + "grad_norm": 0.07162046432495117, + "kept_token_count": 655.059375, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878044664859772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012195533234626054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.25588073730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0589263916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9420, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.1996224358677864, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.12439131140708923, + "grad_norm": 0.07505550980567932, + "kept_token_count": 655.2859375, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878551840782166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012144802417606116, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.6290679931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5884979248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9430, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.2000965490937233, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.12759750708937645, + "grad_norm": 0.06524260342121124, + "kept_token_count": 654.55625, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872308969497681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012769106216728688, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.96966857910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6184844970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9440, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20004646182060243, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.12916994392871856, + "grad_norm": 0.06802783161401749, + "kept_token_count": 654.353125, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.987722110748291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012277895119041204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.9452331542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.51234436035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9450, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.20013777464628218, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.11972588822245597, + "grad_norm": 0.08167137205600739, + "kept_token_count": 654.4375, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9863277018070221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013672303780913353, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.6050170898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3755157470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9460, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20055902749300003, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.12904922664165497, + "grad_norm": 0.06438092142343521, + "kept_token_count": 654.375, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9864348530769348, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01356514384970069, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.50654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9819091796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9470, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19929901659488677, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.12067127823829651, + "grad_norm": 0.061308883130550385, + "kept_token_count": 655.0953125, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870173394680023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012982666958123446, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.06842346191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.26551208496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9480, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.1998631700873375, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.1307532399892807, + "grad_norm": 0.0706627145409584, + "kept_token_count": 654.5828125, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872020840644836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012797925621271133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.31080627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4667602539063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9490, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19994882643222808, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.11693853884935379, + "grad_norm": 0.07830360531806946, + "kept_token_count": 654.6296875, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873290479183197, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012670938204973937, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.772705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41611938476564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9500, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20046068578958512, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.12135626077651977, + "grad_norm": 0.06003063544631004, + "kept_token_count": 654.309375, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9879334807395935, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012066530901938677, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.61012268066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.77935485839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9510, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19934125393629074, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.13084859773516655, + "grad_norm": 0.059776611626148224, + "kept_token_count": 654.8046875, + "learning_rate": 2.451158484961621e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870047807693482, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01299521904438734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.82872314453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9344055175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9520, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.20117487013339996, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.11804978996515274, + "grad_norm": 0.08739439398050308, + "kept_token_count": 653.65625, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867101430892944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01328986007720232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.650048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.17169189453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9530, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19975980669260024, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.12130055204033852, + "grad_norm": 0.06023387610912323, + "kept_token_count": 654.8984375, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.987384033203125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012615968566387891, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.3598388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2371520996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9540, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.19904097318649291, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.12052022144198418, + "grad_norm": 0.05460227280855179, + "kept_token_count": 655.503125, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878979861736298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012102011125534773, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.0744964599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.46290283203126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9550, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19954747408628465, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.11909415572881699, + "grad_norm": 0.06286095082759857, + "kept_token_count": 655.0765625, + "learning_rate": 2.446035097142463e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9869549036026001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013045098353177308, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.17701110839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2362030029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9560, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.19962918162345886, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.12086688801646232, + "grad_norm": 0.0714518204331398, + "kept_token_count": 654.73125, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9875021040439605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01249790657311678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.0705291748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6234466552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9570, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.20047564804553986, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.12737944796681405, + "grad_norm": 0.0700305625796318, + "kept_token_count": 654.51875, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871397256851197, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012860268726944924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.2517608642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15773010253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9580, + "valid_token_count": 818.64375 + }, + { + "actual_token_mask_ratio": 0.20070293098688125, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.1227245420217514, + "grad_norm": 0.06838832050561905, + "kept_token_count": 653.9578125, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9863859236240387, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013614065386354924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.0955383300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49022216796874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20049659460783004, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.12332737371325493, + "grad_norm": 0.07789253443479538, + "kept_token_count": 654.3359375, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9875481963157654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012451811507344247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.8902160644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.02935180664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9600, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19959998577833177, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.13042255416512488, + "grad_norm": 0.050576306879520416, + "kept_token_count": 654.771875, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875809192657471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012419080082327127, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.69325866699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.17783203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9610, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.19968189895153046, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.12659334540367126, + "grad_norm": 0.06841228157281876, + "kept_token_count": 654.8578125, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9863709270954132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013629078585654496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.95362243652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1128265380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9620, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19945267885923385, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.1267838105559349, + "grad_norm": 0.09509670734405518, + "kept_token_count": 655.0421875, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878244638442993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012175520323216915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.31797180175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.61199340820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9630, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.19975224733352662, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.12798884510993958, + "grad_norm": 0.05833921581506729, + "kept_token_count": 654.71875, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878036379814148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012196350656449795, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.8011810302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4706176757812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9640, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.20099884271621704, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.13269369974732398, + "grad_norm": 0.062051475048065186, + "kept_token_count": 653.9203125, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.986712920665741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01328707356005907, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.2656005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.352783203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9650, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.1997341349720955, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.12209525406360626, + "grad_norm": 0.06228942051529884, + "kept_token_count": 655.0796875, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883782625198364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01162174716591835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.8031463623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.69261474609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9660, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.19997602850198745, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.11907877027988434, + "grad_norm": 0.08093085139989853, + "kept_token_count": 654.6828125, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.987716120481491, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012283886689692736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.98041076660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.96575622558595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2008470579981804, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.11775011718273162, + "grad_norm": 0.054924819618463516, + "kept_token_count": 653.9328125, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882983982563018, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011701598297804594, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.11318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4344421386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9680, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20011495649814606, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.1238691620528698, + "grad_norm": 0.062415868043899536, + "kept_token_count": 654.3796875, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876419126987457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012358085811138153, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3918395996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8211303710938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9690, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.1998934715986252, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.1285148046910763, + "grad_norm": 0.07308041304349899, + "kept_token_count": 654.71875, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870320737361908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012967928219586611, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.88096923828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3994445800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9700, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.1995181918144226, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.12068420648574829, + "grad_norm": 0.07227002829313278, + "kept_token_count": 655.0265625, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870494484901429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012950562778860331, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.8296356201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.26441345214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9710, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20007117539644242, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.1264093577861786, + "grad_norm": 0.05204719677567482, + "kept_token_count": 654.6875, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877714931964874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012228490598499776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.98597717285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0896911621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9720, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20002512037754058, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.11656488254666328, + "grad_norm": 0.07837770879268646, + "kept_token_count": 654.6359375, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873354256153106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012664571218192578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.4262390136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1810272216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9730, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20040983110666274, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.12707117795944214, + "grad_norm": 0.08860702812671661, + "kept_token_count": 654.13125, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9869222044944763, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013077796157449483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.28013916015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.14752197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9740, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.20019567906856536, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.1211282528936863, + "grad_norm": 0.06393182277679443, + "kept_token_count": 654.465625, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872846722602844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012715332861989736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.67939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.131494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9750, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20032640546560287, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.13232300207018852, + "grad_norm": 0.0715516060590744, + "kept_token_count": 654.2109375, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868863701820374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013113620411604642, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.40916748046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4403991699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9760, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19947483986616135, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.11490053534507752, + "grad_norm": 0.0721784308552742, + "kept_token_count": 655.2359375, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867623627185822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013237641006708146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.0215698242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3160736083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9770, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.19971508234739305, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.12529306635260581, + "grad_norm": 0.058459434658288956, + "kept_token_count": 654.6390625, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870360612869262, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012963928282260895, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.12294616699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4997131347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9780, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.20027030557394027, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.12562606781721114, + "grad_norm": 0.09903057664632797, + "kept_token_count": 654.4953125, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9860828220844269, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013917184993624687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.23125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1684173583984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9790, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20044276714324952, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.11218047142028809, + "grad_norm": 0.07048171013593674, + "kept_token_count": 654.44375, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.012, + "mgd_cos_sim": 0.988029557466507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011970440205186606, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.35680236816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0370697021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9800, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.1992357313632965, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.11818776205182076, + "grad_norm": 0.07486552000045776, + "kept_token_count": 655.003125, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868898570537568, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013110152911394835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.06431884765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4812469482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9810, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.1999996155500412, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.124285539239645, + "grad_norm": 0.06991161406040192, + "kept_token_count": 654.6109375, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.98667072057724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013329285383224487, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.76128845214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.23797912597655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9820, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2005750820040703, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.12626273408532143, + "grad_norm": 0.05827987566590309, + "kept_token_count": 654.165625, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883094429969788, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01169055150821805, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.75087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8816314697266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.1991759642958641, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.12574406564235688, + "grad_norm": 0.05520467460155487, + "kept_token_count": 655.121875, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873731136322021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01262688459828496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.9199157714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.26690368652345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9840, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.20035457760095596, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.12437828108668328, + "grad_norm": 0.08347604423761368, + "kept_token_count": 653.9828125, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873259246349335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0126740756444633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.5941162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9078002929688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9850, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.1995645746588707, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.12335317656397819, + "grad_norm": 0.06557012349367142, + "kept_token_count": 654.7328125, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9874418318271637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012558163702487945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.7187103271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.780419921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9860, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.2004503458738327, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.13258250802755356, + "grad_norm": 0.0531705766916275, + "kept_token_count": 654.66875, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877192497253418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012280756514519453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.2224395751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5613494873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9870, + "valid_token_count": 818.803125 + }, + { + "actual_token_mask_ratio": 0.20021291077136993, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.12286411598324776, + "grad_norm": 0.0910646840929985, + "kept_token_count": 654.43125, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873099148273468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012690085545182228, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.2260681152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9011566162109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9880, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.1991576984524727, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.12740612179040908, + "grad_norm": 0.0770474374294281, + "kept_token_count": 655.3375, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9871603608131408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01283963704481721, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.40472412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0309814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9890, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19928720742464065, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.12054401710629463, + "grad_norm": 0.07226181030273438, + "kept_token_count": 655.146875, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878653764724732, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01213461896404624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.71644592285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6805480957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9900, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19971440583467484, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.12367491647601128, + "grad_norm": 0.059866346418857574, + "kept_token_count": 655.0703125, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877489924430847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012251021154224873, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.99786682128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.77210998535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9910, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.19908679872751237, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.12694723084568976, + "grad_norm": 0.06827345490455627, + "kept_token_count": 655.2015625, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873948156833648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01260518915951252, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.62738037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11424865722654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9920, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.19950196146965027, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.13327767476439475, + "grad_norm": 0.08965514600276947, + "kept_token_count": 655.0640625, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875978648662567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012402132898569108, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.22747802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5518859863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9930, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19949337393045424, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.12792170643806458, + "grad_norm": 0.06892789155244827, + "kept_token_count": 655.16875, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882266342639923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011773374304175377, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.2405059814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.18165588378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9940, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.19975237995386125, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.12439488992094994, + "grad_norm": 0.05920809879899025, + "kept_token_count": 654.6703125, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881023228168487, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011897681560367347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.3435455322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.742431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.2003598317503929, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.12226147428154946, + "grad_norm": 0.05875271186232567, + "kept_token_count": 654.24375, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872316837310791, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01276832539588213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.10755920410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2394653320313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9960, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.2001565635204315, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.13176939338445665, + "grad_norm": 0.06337963044643402, + "kept_token_count": 654.6875, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9861524701118469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013847545348107814, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.44956970214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8710205078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9970, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.20121267884969712, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.11513479128479957, + "grad_norm": 0.06074368208646774, + "kept_token_count": 653.890625, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.987326693534851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012673310469835997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.7105987548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.90953674316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9980, + "valid_token_count": 818.603125 + }, + { + "actual_token_mask_ratio": 0.19968001991510392, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.12063644900918007, + "grad_norm": 0.06070453301072121, + "kept_token_count": 654.878125, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9879604279994965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012039568368345499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.09771423339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7622100830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9990, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20029274374246597, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.12722809463739396, + "grad_norm": 0.05423595383763313, + "kept_token_count": 654.259375, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9875229477882386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012477061990648509, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.6615844726563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.15955505371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10000, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20019359290599822, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.1170505665242672, + "grad_norm": 0.06502808630466461, + "kept_token_count": 654.6484375, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873198449611664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012680148892104625, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.5271514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5110626220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10010, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.20067771524190903, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.1233773909509182, + "grad_norm": 0.06672415137290955, + "kept_token_count": 654.1203125, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880659520626068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01193406516686082, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.47478942871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9916748046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10020, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.20041294693946837, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.1224337637424469, + "grad_norm": 0.05858530104160309, + "kept_token_count": 654.5765625, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876051783561707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012394824158400297, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.8203430175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2884246826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10030, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.2004382774233818, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.11806916892528534, + "grad_norm": 0.07608159631490707, + "kept_token_count": 654.178125, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878549337387085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012145061697810888, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.7394195556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0935913085938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19848856031894685, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.12533625066280366, + "grad_norm": 0.07563424110412598, + "kept_token_count": 655.7421875, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878996074199676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012100392766296864, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.8202209472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4139373779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19953189492225648, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.11611036881804467, + "grad_norm": 0.05994318053126335, + "kept_token_count": 655.0578125, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873880565166473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012611936964094638, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.4912414550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7842132568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10060, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.1999606594443321, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.1282789573073387, + "grad_norm": 0.06695005297660828, + "kept_token_count": 654.7, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866478741168976, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013352133706212044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.8904266357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9990692138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10070, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20032535642385482, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.12235981747508048, + "grad_norm": 0.06451942771673203, + "kept_token_count": 654.1390625, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870935499668121, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012906457856297493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.1031951904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7599670410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10080, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.19944326281547547, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.11510377824306488, + "grad_norm": 0.06874044239521027, + "kept_token_count": 655.009375, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.987225878238678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012774139735847711, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.39056091308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.31981506347654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10090, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20060164779424666, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.12579077258706092, + "grad_norm": 0.06427202373743057, + "kept_token_count": 654.23125, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884234726428985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011576525866985321, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.3589752197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.83449096679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2007445603609085, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.13449498638510704, + "grad_norm": 0.08365695178508759, + "kept_token_count": 653.796875, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9867937982082366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013206207193434238, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.4120178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5785369873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10110, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.2000977262854576, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.12444196715950966, + "grad_norm": 0.06277857720851898, + "kept_token_count": 654.61875, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875988900661469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012401100806891918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.2642852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1244812011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10120, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.199974949657917, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.11588009670376778, + "grad_norm": 0.059996459633111954, + "kept_token_count": 654.4234375, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885065376758575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011493450589478016, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.1193389892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0501770019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10130, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.2005816489458084, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.12135959342122078, + "grad_norm": 0.05533086508512497, + "kept_token_count": 654.5171875, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9874878406524659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012512159906327725, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.0877349853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2899871826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10140, + "valid_token_count": 818.7421875 + }, + { + "actual_token_mask_ratio": 0.20013115108013152, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.1155388742685318, + "grad_norm": 0.059317950159311295, + "kept_token_count": 654.546875, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877504348754883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012249570991843938, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.54619750976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6586090087891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10150, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.1998257040977478, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.11993077322840691, + "grad_norm": 0.08188671618700027, + "kept_token_count": 654.7046875, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870950818061829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012904917262494564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.25579528808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.63915100097654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10160, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.20011563301086427, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.1347632311284542, + "grad_norm": 0.06598702818155289, + "kept_token_count": 654.64375, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880335330963135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011966456472873688, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.33872375488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.64073486328124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10170, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19982240498065948, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.13041911348700524, + "grad_norm": 0.07564643770456314, + "kept_token_count": 654.7453125, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871440231800079, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012855984829366207, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.5417022705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05906372070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10180, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19961064755916597, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.12257490158081055, + "grad_norm": 0.05589563772082329, + "kept_token_count": 654.915625, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882855415344238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011714451480656862, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.69671630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5057342529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10190, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20023862570524215, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.1197464182972908, + "grad_norm": 0.05849594995379448, + "kept_token_count": 654.65625, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873460114002228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012653970066457987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.0216033935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0936676025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10200, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.2007174536585808, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.12577513977885246, + "grad_norm": 0.05620235949754715, + "kept_token_count": 654.1078125, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881795823574067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011820410657674075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.0491058349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2846405029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10210, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.19999000430107117, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.11929953321814538, + "grad_norm": 0.05678541585803032, + "kept_token_count": 654.5953125, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881630361080169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011836974509060382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.16013488769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5233947753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10220, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20038362592458725, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.10989074409008026, + "grad_norm": 0.06535451859235764, + "kept_token_count": 654.0796875, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884467422962189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01155326422303915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.91196899414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6098266601563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10230, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.20011062771081925, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.11739064455032348, + "grad_norm": 0.06373503804206848, + "kept_token_count": 654.6296875, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870879828929902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012912012916058303, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.721142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3516052246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20004518330097198, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.11676414385437965, + "grad_norm": 0.0830957219004631, + "kept_token_count": 654.4875, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9871068716049194, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012893123924732209, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.9119598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.64308471679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10250, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20082085877656936, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.12611312344670295, + "grad_norm": 0.06087326258420944, + "kept_token_count": 653.903125, + "learning_rate": 2.353488759518022e-05, + "loss": 0.012, + "mgd_cos_sim": 0.98800288438797, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011997120827436447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.69701232910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6924194335937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10260, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.20077328234910966, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.13601802811026573, + "grad_norm": 0.04870690405368805, + "kept_token_count": 653.9671875, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884998381137848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011500170081853866, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0169616699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7480712890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10270, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2002411365509033, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.1215272717177868, + "grad_norm": 0.07618968933820724, + "kept_token_count": 654.1390625, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9868321359157562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013167866133153438, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.13645629882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0570587158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10280, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.20078605711460112, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.11886890232563019, + "grad_norm": 0.07147389650344849, + "kept_token_count": 653.8421875, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880254924297333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011974516697227955, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.1335113525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.79713439941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10290, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.19982339292764664, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.11949122622609139, + "grad_norm": 0.06317493319511414, + "kept_token_count": 654.8125, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883437991142273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011656202748417855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.0259124755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.42577209472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10300, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.1992842212319374, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.11870416477322579, + "grad_norm": 0.08305469900369644, + "kept_token_count": 655.0625, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9868678331375123, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013132178597152233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.32789306640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.24204711914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10310, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20056204050779342, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.11971570253372192, + "grad_norm": 0.0664701759815216, + "kept_token_count": 654.0171875, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873283565044403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012671652901917695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.26926879882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.69719848632815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10320, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20021220296621323, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.11590229943394662, + "grad_norm": 0.05545041710138321, + "kept_token_count": 654.5609375, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872193038463593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01278070518746972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.6542999267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.97227172851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10330, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2002855658531189, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.11951241567730904, + "grad_norm": 0.07046525925397873, + "kept_token_count": 654.0421875, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870904862880707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01290948623791337, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.80137634277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.065234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10340, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.1995686873793602, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.1282110407948494, + "grad_norm": 0.057043708860874176, + "kept_token_count": 654.9046875, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872502088546753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012749794404953718, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.2086181640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0068634033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10350, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.19983442425727843, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.12691968008875848, + "grad_norm": 0.05581511929631233, + "kept_token_count": 654.940625, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9876736462116241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012326361984014511, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.1265045166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.92258605957034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10360, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.20026267915964127, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.11820420175790787, + "grad_norm": 0.05286063998937607, + "kept_token_count": 654.6484375, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880711734294891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011928832065314055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.7226531982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2014526367187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10370, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.20037283301353453, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.12247274741530419, + "grad_norm": 0.06605438143014908, + "kept_token_count": 654.334375, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870512127876282, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012948779482394457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.9930145263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2999542236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10380, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.20025548040866853, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.12548713833093644, + "grad_norm": 0.05816427990794182, + "kept_token_count": 654.4046875, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882750511169434, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011724947020411491, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.8774932861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0399230957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10390, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19962144494056702, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.1223677210509777, + "grad_norm": 0.05393887683749199, + "kept_token_count": 655.2703125, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880510091781616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011949006281793118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.57310485839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.077197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10400, + "valid_token_count": 818.6984375 + }, + { + "actual_token_mask_ratio": 0.2003021478652954, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.1208829089999199, + "grad_norm": 0.05703045427799225, + "kept_token_count": 654.38125, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9875108361244201, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012489168532192707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.88660583496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.12876586914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10410, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20030897706747056, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.1175291433930397, + "grad_norm": 0.06030404940247536, + "kept_token_count": 654.175, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9867262125015259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01327379560098052, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.48059997558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4729705810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10420, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.20034399926662444, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.1175020083785057, + "grad_norm": 0.07695195823907852, + "kept_token_count": 654.3640625, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876078009605408, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012392197083681823, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.0124237060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.91766052246095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10430, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.198556949198246, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.1196998491883278, + "grad_norm": 0.07067466527223587, + "kept_token_count": 655.83125, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876152575016022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012384743615984916, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0957489013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.71746215820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10440, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.19989925473928452, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.122008016705513, + "grad_norm": 0.08075716346502304, + "kept_token_count": 654.578125, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.987276428937912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01272357553243637, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.20325622558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.17164611816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10450, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20001475661993026, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.12005056962370872, + "grad_norm": 0.08401428163051605, + "kept_token_count": 654.4140625, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.988630759716034, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01136924009770155, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.20335693359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.01878051757814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10460, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.19966465532779692, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.11346558034420014, + "grad_norm": 0.06616965681314468, + "kept_token_count": 654.9, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.988399863243103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01160013247281313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.64765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5713195800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10470, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.19912881702184676, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.11956460028886795, + "grad_norm": 0.06588380038738251, + "kept_token_count": 655.715625, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9879350960254669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012064907047897578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.7123168945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3662841796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10480, + "valid_token_count": 818.753125 + }, + { + "actual_token_mask_ratio": 0.19915010929107665, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.12851888611912726, + "grad_norm": 0.07337218523025513, + "kept_token_count": 655.259375, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870280385017395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012971973605453968, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3615997314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.47751770019534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20040818601846694, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.1389896646142006, + "grad_norm": 0.0687904953956604, + "kept_token_count": 654.2921875, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875925481319427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012407452799379826, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.2131744384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.94895629882814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10500, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20082181245088576, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.11867896169424057, + "grad_norm": 0.07506611198186874, + "kept_token_count": 653.7421875, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9872182428836822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012781756650656461, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.54310913085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07729187011716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10510, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.1990290254354477, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.12473684400320054, + "grad_norm": 0.06253915280103683, + "kept_token_count": 655.521875, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9879614770412445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012038512248545885, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.0818389892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6253662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10520, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2000571981072426, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.1276077412068844, + "grad_norm": 0.06834550946950912, + "kept_token_count": 654.6015625, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9874493479728699, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012550663016736508, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.50263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.62750854492185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10530, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19996484518051147, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.1256166160106659, + "grad_norm": 0.04662946239113808, + "kept_token_count": 654.4984375, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877682566642761, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012231753952801228, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.77341613769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.16859741210936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10540, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2006152555346489, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.1255800060927868, + "grad_norm": 0.060293424874544144, + "kept_token_count": 653.83125, + "learning_rate": 2.313627577952336e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870227396488189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012977258954197168, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.89853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.51357421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10550, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.19989814162254332, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.12747765854001045, + "grad_norm": 0.06347764283418655, + "kept_token_count": 654.55625, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881482005119324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011851792596280575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.10853271484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.39601745605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10560, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.1995253548026085, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.12333922609686851, + "grad_norm": 0.08433374017477036, + "kept_token_count": 655.1984375, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9873722732067108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012627732940018177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.1051971435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84473876953126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10570, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.19999125152826308, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.11306222155690193, + "grad_norm": 0.06179918721318245, + "kept_token_count": 654.346875, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881680369377136, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011831981502473355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.59840087890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5602966308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10580, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.19974106997251512, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.1285833701491356, + "grad_norm": 0.0677461177110672, + "kept_token_count": 654.834375, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0128, + "mgd_cos_sim": 0.9871573805809021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012842603400349617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4903259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8966461181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10590, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.1985154926776886, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.12019820734858513, + "grad_norm": 0.058465130627155304, + "kept_token_count": 656.0453125, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880930542945862, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011906934902071953, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.6518615722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.72608642578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10600, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.20037501007318498, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.12265941053628922, + "grad_norm": 0.05303278565406799, + "kept_token_count": 654.60625, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.98819580078125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011804189719259738, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.86591796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07312927246096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10610, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20163936913013458, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.14089274033904076, + "grad_norm": 0.059030722826719284, + "kept_token_count": 653.35, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877322971820831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012267687171697617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.9737243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.22121276855466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10620, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20076287537813187, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.13442656621336938, + "grad_norm": 0.06472405046224594, + "kept_token_count": 653.809375, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877319574356079, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012268052808940411, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.6472564697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8658020019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10630, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20050610601902008, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.12684944048523902, + "grad_norm": 0.05217206850647926, + "kept_token_count": 654.259375, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9875991404056549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012400842178612948, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.0461944580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.60841064453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10640, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.20060383379459382, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.1301615945994854, + "grad_norm": 0.053759410977363586, + "kept_token_count": 654.1875, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.987905865907669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012094138562679291, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.8152221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4705047607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10650, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20066096633672714, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.10873325690627098, + "grad_norm": 0.06959430873394012, + "kept_token_count": 653.984375, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881095886230469, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011890406534075738, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.2673675537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5246337890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10660, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20000975131988524, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.12487560585141182, + "grad_norm": 0.056604571640491486, + "kept_token_count": 654.4140625, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.988301819562912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011698172148317098, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.7012023925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87919921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10670, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.20001111924648285, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.13304931446909904, + "grad_norm": 0.07475342601537704, + "kept_token_count": 654.5171875, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877903938293457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012209606543183326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.6352874755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.55762634277346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10680, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20020344406366347, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.12183583080768585, + "grad_norm": 0.0643640011548996, + "kept_token_count": 654.81875, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.987055504322052, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012944508157670499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.2907653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.28148193359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10690, + "valid_token_count": 818.728125 + }, + { + "actual_token_mask_ratio": 0.20005978941917418, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.12475684136152268, + "grad_norm": 0.05398629605770111, + "kept_token_count": 654.6875, + "learning_rate": 2.292680897767441e-05, + "loss": 0.012, + "mgd_cos_sim": 0.98802290558815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011977076437324286, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.06597290039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7375152587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10700, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.199513278901577, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.12378698810935021, + "grad_norm": 0.05757468566298485, + "kept_token_count": 654.9125, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876394093036651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012360596470534802, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.5386169433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7554046630859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10710, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20016134530305862, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.12610676810145377, + "grad_norm": 0.05425717309117317, + "kept_token_count": 654.2015625, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.98823122382164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011768774315714835, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.1817932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.63897705078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10720, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.19932931661605835, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.12467588186264038, + "grad_norm": 0.08215998858213425, + "kept_token_count": 655.0015625, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878243327140808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012175661697983741, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.0830505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4010528564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10730, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.1995003417134285, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.12269532084465026, + "grad_norm": 0.056925978511571884, + "kept_token_count": 654.9625, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882945060729981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011705491971224547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.0597290039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5624145507812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10740, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20074382424354553, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.12677389979362488, + "grad_norm": 0.052771009504795074, + "kept_token_count": 654.0453125, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882081389427185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011791853420436382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.1166046142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2896667480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10750, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.19940887987613679, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.12440383657813073, + "grad_norm": 0.0626675933599472, + "kept_token_count": 654.978125, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878101229667664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012189872562885284, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.98508911132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1274658203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10760, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.19967273026704788, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.11258962377905846, + "grad_norm": 0.05178195610642433, + "kept_token_count": 654.625, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880075633525849, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011992454994469881, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.53074951171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9647613525391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10770, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.19972336292266846, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.11342804804444313, + "grad_norm": 0.0591096505522728, + "kept_token_count": 655.0046875, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884077668190002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011592227779328824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.54652404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2554016113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19986535161733626, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.11583126410841942, + "grad_norm": 0.05644712224602699, + "kept_token_count": 654.6734375, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887988269329071, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01120117623358965, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.7572326660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.71343688964845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10790, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20076136738061906, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.13609059229493142, + "grad_norm": 0.07325281947851181, + "kept_token_count": 654.0484375, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9864791870117188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01352082509547472, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.87506408691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4028259277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.1992545858025551, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.12601775228977202, + "grad_norm": 0.06751612573862076, + "kept_token_count": 655.3515625, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.988164484500885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011835503298789263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.1817138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78057556152345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10810, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.199461829662323, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.12552915439009665, + "grad_norm": 0.051213521510362625, + "kept_token_count": 655.01875, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881467461585999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011853259708732366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.7800628662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0788604736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10820, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20133306533098222, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.13033057227730752, + "grad_norm": 0.07277359068393707, + "kept_token_count": 653.7890625, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884596347808838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01154035869985819, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4353729248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.48968811035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10830, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.19953529983758928, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.12650531753897667, + "grad_norm": 0.05931708216667175, + "kept_token_count": 655.08125, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9875389039516449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012461097352206707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.41290588378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1803771972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10840, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20077624320983886, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.12367537841200829, + "grad_norm": 0.0683259442448616, + "kept_token_count": 654.115625, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883142292499543, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011685780901461839, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.4961761474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.60437927246096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10850, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.2003497987985611, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.12325512394309043, + "grad_norm": 0.07390391081571579, + "kept_token_count": 654.45625, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880531847476959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011946829501539469, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0345489501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.67345275878904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10860, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.19976220279932022, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.11689618751406669, + "grad_norm": 0.05868269130587578, + "kept_token_count": 654.8875, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880971074104309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01190289929509163, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.15125427246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.00322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10870, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20007942467927933, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.12892681881785392, + "grad_norm": 0.06267870217561722, + "kept_token_count": 654.775, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881911933422088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011808801721781493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.5270263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.37143859863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10880, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.20038454234600067, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.11527860760688782, + "grad_norm": 0.06079137325286865, + "kept_token_count": 654.396875, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.987702465057373, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012297532428056002, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.54770812988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.02296142578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.19929229319095612, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.11151496842503547, + "grad_norm": 0.0744931772351265, + "kept_token_count": 655.1671875, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878424346446991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012157553620636464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.60602416992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8802001953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10900, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20056854635477067, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.12477808073163033, + "grad_norm": 0.05702215060591698, + "kept_token_count": 654.0265625, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880631148815155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011936878599226475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.06333618164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.22276611328124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.1998745545744896, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.12063800990581512, + "grad_norm": 0.061832182109355927, + "kept_token_count": 654.91875, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882836222648621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011716368235647678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.0551300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5596984863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10920, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19918845295906068, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.12429860457777978, + "grad_norm": 0.06608444452285767, + "kept_token_count": 655.296875, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9870882332324982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012911771982908249, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.50361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.698193359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10930, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2003024026751518, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.12769161015748978, + "grad_norm": 0.05304216593503952, + "kept_token_count": 654.6625, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887115180492401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011288494430482387, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8220733642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0124237060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10940, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.20019602924585342, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.12243214696645736, + "grad_norm": 0.06504034996032715, + "kept_token_count": 654.5625, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873103201389313, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012689684610813856, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.1399291992187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.71360473632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10950, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.19973860383033754, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.1170061133801937, + "grad_norm": 0.052807312458753586, + "kept_token_count": 655.065625, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9865596652030945, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0134403420612216, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.43650512695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4756103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10960, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.20034506320953369, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.12979274839162827, + "grad_norm": 0.05788453295826912, + "kept_token_count": 654.3234375, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884562849998474, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011543720308691264, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.0464111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05816040039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10970, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20042723119258882, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.10861792117357254, + "grad_norm": 0.10585547983646393, + "kept_token_count": 654.2890625, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878219366073608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012178070656955243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.81944274902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.32378540039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.19973374009132386, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.1218987375497818, + "grad_norm": 0.07596234977245331, + "kept_token_count": 654.8765625, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877124309539795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012287571094930172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.2146881103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3928680419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.20029022097587584, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.11878496259450913, + "grad_norm": 0.08697480708360672, + "kept_token_count": 654.4359375, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878043472766876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012195645645260812, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.982177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5562225341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11000, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20041783452033995, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.1262476585805416, + "grad_norm": 0.05654563382267952, + "kept_token_count": 654.215625, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882284343242645, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011771561484783888, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.4973541259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8792236328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11010, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19999164789915086, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.1273021452128887, + "grad_norm": 0.05776049941778183, + "kept_token_count": 654.7890625, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881137728691101, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011886222753673792, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.59697875976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.18227233886716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11020, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.20011833906173707, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.11877478659152985, + "grad_norm": 0.05298857390880585, + "kept_token_count": 654.5265625, + "learning_rate": 2.245843529059e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878686666488647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012131329346448182, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.83779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4599578857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11030, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19910681247711182, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.12034593522548676, + "grad_norm": 0.07588884234428406, + "kept_token_count": 655.3546875, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872829735279083, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012717035319656133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.6648376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.08818969726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11040, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.1997393399477005, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.1302987091243267, + "grad_norm": 0.05329250916838646, + "kept_token_count": 654.825, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9874886631965637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012511330377310514, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.6370147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.875244140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11050, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.1992822751402855, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.11618089005351066, + "grad_norm": 0.060130588710308075, + "kept_token_count": 655.4390625, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877703070640564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01222969489172101, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.29442443847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15221252441404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11060, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.19973569810390474, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.12331553474068642, + "grad_norm": 0.061078645288944244, + "kept_token_count": 654.7578125, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9879199385643005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012080066278576851, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.5614349365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4660675048828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11070, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19967023581266402, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.12241555154323577, + "grad_norm": 0.058393388986587524, + "kept_token_count": 654.696875, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887418210506439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011258175317198037, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.94391479492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7395904541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11080, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.20072828531265258, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.12558164522051812, + "grad_norm": 0.059467729181051254, + "kept_token_count": 653.93125, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881502270698548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01184975765645504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.79741821289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8285736083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11090, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20033476054668425, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.12576504573225974, + "grad_norm": 0.04965083301067352, + "kept_token_count": 654.084375, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886783719062805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011321630887687206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6824005126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.00208435058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11100, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.2005688354372978, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.11586533188819885, + "grad_norm": 0.047491319477558136, + "kept_token_count": 654.0609375, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.988492465019226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011507522221654654, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.7760284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5932861328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11110, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.19983592480421067, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.126823890209198, + "grad_norm": 0.05940958112478256, + "kept_token_count": 654.715625, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885826528072357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011417342629283667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.3875732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5422332763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11120, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20088325291872025, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.13863285556435584, + "grad_norm": 0.047049012035131454, + "kept_token_count": 653.628125, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870265007019043, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012973477598279715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.61965026855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6685821533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11130, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.19970270097255707, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.1216955915093422, + "grad_norm": 0.061791010200977325, + "kept_token_count": 654.7171875, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.988869845867157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011130163911730051, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8356140136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0343719482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11140, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.20007904171943663, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.11659352406859398, + "grad_norm": 0.05850651487708092, + "kept_token_count": 654.5484375, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888577044010163, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011142283119261265, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.0390258789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.99204711914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11150, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19960561394691467, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.12996637523174287, + "grad_norm": 0.06812908500432968, + "kept_token_count": 654.9203125, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886398255825043, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011360164731740952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.50096130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5074096679688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11160, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.2004536658525467, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.12378147318959236, + "grad_norm": 0.05421586334705353, + "kept_token_count": 653.9578125, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886361718177795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011363841220736503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.72963256835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.22027282714845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11170, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.20031987726688386, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.11840123161673546, + "grad_norm": 0.06848160922527313, + "kept_token_count": 654.3140625, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881553828716279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011844596732407809, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.174072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6237091064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11180, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20040929168462754, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.13463197499513627, + "grad_norm": 0.06097722053527832, + "kept_token_count": 654.2796875, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9876649498939514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012335052248090506, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.9904388427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9851837158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11190, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20059392154216765, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.12066942602396011, + "grad_norm": 0.06901080906391144, + "kept_token_count": 654.075, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878564476966858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012143542803823949, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0113891601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9882110595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20087756514549254, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.12226707935333252, + "grad_norm": 0.08928624540567398, + "kept_token_count": 653.71875, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881809651851654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01181904161348939, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.0488342285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4302001953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11210, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.20085848420858382, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.10671475902199745, + "grad_norm": 0.054007481783628464, + "kept_token_count": 653.934375, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888978362083435, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011102153733372689, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0302398681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1603118896484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11220, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2003715991973877, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.12107821479439736, + "grad_norm": 0.08293504267930984, + "kept_token_count": 653.990625, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877666294574737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012233373895287513, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.118310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.95769958496095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11230, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.19960809946060182, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.12472402602434159, + "grad_norm": 0.04991695284843445, + "kept_token_count": 654.740625, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9866042971611023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013395680952817202, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.04139099121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8087219238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11240, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.19905790239572524, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.1269457831978798, + "grad_norm": 0.05486002936959267, + "kept_token_count": 655.0703125, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9872602164745331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012739784270524978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.69765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8332489013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11250, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.2000888243317604, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.11730986461043358, + "grad_norm": 0.06376483291387558, + "kept_token_count": 654.6265625, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878659069538116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012134082894772292, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.14796142578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1346801757812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11260, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.19967576563358308, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.12381386980414391, + "grad_norm": 0.06752503663301468, + "kept_token_count": 654.8015625, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883368194103241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011663173325359821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.85669860839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5552459716797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11270, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2014367938041687, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.11731600761413574, + "grad_norm": 0.05703216418623924, + "kept_token_count": 653.721875, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.988523268699646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011476731859147549, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.07490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3861053466797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11280, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.20001179575920106, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.11881067454814911, + "grad_norm": 0.07004217803478241, + "kept_token_count": 654.6625, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889240622520447, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011075931042432785, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.3537231445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7098327636719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11290, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2002854749560356, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.11440427675843239, + "grad_norm": 0.051838699728250504, + "kept_token_count": 654.5046875, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.012, + "mgd_cos_sim": 0.988018810749054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011981192044913769, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.64756774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0108154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11300, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.20017769932746887, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.11681581437587737, + "grad_norm": 0.05445314198732376, + "kept_token_count": 654.7453125, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878009736537934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012199033424258232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.930517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63878479003904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11310, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.19945873320102692, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.11654655188322068, + "grad_norm": 0.04761010780930519, + "kept_token_count": 655.21875, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.988260817527771, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01173917856067419, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.66412353515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0336456298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11320, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.19961130917072295, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.12280776351690292, + "grad_norm": 0.049234259873628616, + "kept_token_count": 655.43125, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887526631355286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011247328761965037, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.5439056396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.064453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11330, + "valid_token_count": 818.890625 + }, + { + "actual_token_mask_ratio": 0.19994528144598006, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.1240171879529953, + "grad_norm": 0.07000074535608292, + "kept_token_count": 654.4984375, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9868244647979736, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013175525143742562, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.407666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3470855712891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11340, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.200733183324337, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.12709306925535202, + "grad_norm": 0.05096947029232979, + "kept_token_count": 653.9609375, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884909391403198, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011509049031883479, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.40557861328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15085754394534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11350, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.20057480335235595, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.11735972464084625, + "grad_norm": 0.06935510039329529, + "kept_token_count": 654.103125, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9866556286811828, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.013344364427030086, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.4848236083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.26630249023435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11360, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19945013225078584, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.12040538117289543, + "grad_norm": 0.0526849590241909, + "kept_token_count": 655.475, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881678938865661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011832108162343502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.7527099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11370, + "valid_token_count": 818.778125 + }, + { + "actual_token_mask_ratio": 0.20118749737739564, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.12734218165278435, + "grad_norm": 0.058681439608335495, + "kept_token_count": 653.7828125, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883653700351716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011634633503854274, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7041870117188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5889892578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11380, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.19995470046997071, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.12246930077672005, + "grad_norm": 0.0623837411403656, + "kept_token_count": 654.8, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880979061126709, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011902075074613095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.92421569824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4563262939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.19970571249723434, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.12946696802973748, + "grad_norm": 0.05132851004600525, + "kept_token_count": 654.9390625, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889757990837097, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011024196725338697, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.0792205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2723815917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11400, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.19975431710481645, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.11936464458703995, + "grad_norm": 0.0639556348323822, + "kept_token_count": 655.0828125, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877499163150787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012250095326453447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.6187347412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.307421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11410, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.1986571416258812, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.1174478754401207, + "grad_norm": 0.07761666923761368, + "kept_token_count": 655.671875, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884791672229767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011520820669829845, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.33607482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9268798828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11420, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19962854385375978, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.12054232433438301, + "grad_norm": 0.05344405397772789, + "kept_token_count": 654.6703125, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893153965473175, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010684617795050144, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.26844482421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9693267822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11430, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.1992887958884239, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.10957752019166947, + "grad_norm": 0.08328820765018463, + "kept_token_count": 655.3984375, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887867510318756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01121326070278883, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.5594116210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7992767333984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11440, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.2006702870130539, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.12431685253977776, + "grad_norm": 0.05330657586455345, + "kept_token_count": 653.89375, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886779069900513, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011322085838764906, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.2955810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.94830932617185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11450, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20039364397525788, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.12099391892552376, + "grad_norm": 0.07111543416976929, + "kept_token_count": 654.3125, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876487255096436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012351257912814616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.10422973632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.67710876464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.199782495200634, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.11856336072087288, + "grad_norm": 0.07219228893518448, + "kept_token_count": 654.9328125, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886798441410065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011320146452635526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.5376861572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9561065673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11470, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.19912007302045823, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.11999384909868241, + "grad_norm": 0.05385151505470276, + "kept_token_count": 655.0265625, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886800885200501, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011319913435727358, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.8528625488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.97608337402346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11480, + "valid_token_count": 817.8890625 + }, + { + "actual_token_mask_ratio": 0.19922388195991517, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.12171332016587258, + "grad_norm": 0.06158357486128807, + "kept_token_count": 655.340625, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885560631752014, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011443914752453566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.9879669189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.33983459472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11490, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20118208974599838, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.12264842838048935, + "grad_norm": 0.0613243505358696, + "kept_token_count": 653.609375, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880281925201416, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011971820425242186, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.2092346191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5313232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11500, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20011971592903138, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.11393495425581932, + "grad_norm": 0.07337246090173721, + "kept_token_count": 654.515625, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884777724742889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01152221318334341, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9104736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8257202148437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11510, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2002359464764595, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.12980401664972305, + "grad_norm": 0.07713761180639267, + "kept_token_count": 654.5484375, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0124, + "mgd_cos_sim": 0.9876291275024414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012370866350829601, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.6888397216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2696838378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11520, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20014841258525848, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.12990734875202178, + "grad_norm": 0.05696501582860947, + "kept_token_count": 654.5125, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886938691139221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01130613572895527, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.10480346679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.76584777832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11530, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20041225850582123, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.12417783439159394, + "grad_norm": 0.11363585293292999, + "kept_token_count": 654.3921875, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882594227790833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011740575078874826, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.3609283447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.57696533203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11540, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20034365206956864, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.13165437430143356, + "grad_norm": 0.05012456327676773, + "kept_token_count": 654.38125, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885477721691132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011452208925038575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.7823913574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.49888610839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11550, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19985740780830383, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.12255942150950432, + "grad_norm": 0.05262032896280289, + "kept_token_count": 655.0109375, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888493835926055, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01115060606971383, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7832550048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3867950439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11560, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.20128704309463502, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.12450996115803718, + "grad_norm": 0.07300002127885818, + "kept_token_count": 653.5984375, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887398719787598, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01126011684536934, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.6955993652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8344665527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11570, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.20118762105703353, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.12382777482271194, + "grad_norm": 0.05050879344344139, + "kept_token_count": 653.5640625, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889220595359802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011077925842255355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.8125732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.302197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11580, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.199267914891243, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.12968022227287293, + "grad_norm": 0.05741307511925697, + "kept_token_count": 655.0421875, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890735864639282, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01092640608549118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.36455688476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2794677734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11590, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20072106570005416, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.12014998197555542, + "grad_norm": 0.066231369972229, + "kept_token_count": 654.1984375, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885812044143677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011418784130364656, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.2858154296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22223205566405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11600, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.20049462616443633, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.13605317398905753, + "grad_norm": 0.06799997389316559, + "kept_token_count": 654.5, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877561032772064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01224389923736453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.532763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5880889892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11610, + "valid_token_count": 818.63125 + }, + { + "actual_token_mask_ratio": 0.19977030009031296, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.12760234177112578, + "grad_norm": 0.05880764126777649, + "kept_token_count": 654.85625, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9873488128185273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012651183828711509, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.7088134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5120452880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11620, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.1999183863401413, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.12909945398569106, + "grad_norm": 0.06885053962469101, + "kept_token_count": 654.7046875, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880590379238129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011940956860780717, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.0163330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0021911621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20065265148878098, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.12736035138368607, + "grad_norm": 0.07199438661336899, + "kept_token_count": 653.9734375, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9891223430633544, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010877661872655153, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.38104858398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66522521972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11640, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.20056697726249695, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.1234574519097805, + "grad_norm": 0.05298162251710892, + "kept_token_count": 654.315625, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890693604946137, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010930644068866967, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.2886199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.75022583007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11650, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.19946832507848739, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.122872094810009, + "grad_norm": 0.05488177016377449, + "kept_token_count": 655.0921875, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9890031158924103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010996870044618846, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.35797729492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5456878662109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11660, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.20040960162878035, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.13039517849683763, + "grad_norm": 0.04947544261813164, + "kept_token_count": 654.2234375, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887308061122895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01126918252557516, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.83587951660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2490753173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11670, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20068853199481965, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.11891376003623008, + "grad_norm": 0.051757264882326126, + "kept_token_count": 654.034375, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882101058959961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011789895594120026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.9509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.75409240722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11680, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2008750006556511, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.12936004996299744, + "grad_norm": 0.051347315311431885, + "kept_token_count": 653.95625, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889921069145202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0110078826546669, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.29054565429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.49314880371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20044503957033158, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.11724196895956993, + "grad_norm": 0.05599705129861832, + "kept_token_count": 654.1921875, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883123815059662, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011687615793198347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.3531494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.68581237792966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11700, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.19923235923051835, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.12891100943088532, + "grad_norm": 0.05802154168486595, + "kept_token_count": 655.0171875, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883892655372619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011610749550163746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.54778442382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5038299560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11710, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.1999657392501831, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.11476888731122017, + "grad_norm": 0.05947589501738548, + "kept_token_count": 654.55, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884923696517944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011507632024586201, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.33095397949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.998291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11720, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.19962205588817597, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.12426308318972587, + "grad_norm": 0.07233915477991104, + "kept_token_count": 655.053125, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882105588912964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011789432726800442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.7905548095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8587280273438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11730, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.19984547942876815, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.12068067789077759, + "grad_norm": 0.06210342422127724, + "kept_token_count": 654.809375, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882496237754822, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011750372126698494, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.1791748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2497924804687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11740, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19841219633817672, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.12946399301290512, + "grad_norm": 0.05928671360015869, + "kept_token_count": 655.8625, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.988225269317627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011774744838476181, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.0091613769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.10426635742186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11750, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19997439086437224, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.11799177974462509, + "grad_norm": 0.05721936374902725, + "kept_token_count": 654.753125, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881446897983551, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011855311039835215, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.2840270996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.825634765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11760, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20088271498680116, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.12403687238693237, + "grad_norm": 0.049161095172166824, + "kept_token_count": 653.8546875, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889690697193145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011030928976833821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.14397583007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0548614501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11770, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20077026188373565, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.11735004186630249, + "grad_norm": 0.04993841424584389, + "kept_token_count": 654.1796875, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.989251971244812, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010748026333749295, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.23253173828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.65077209472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.2000689446926117, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.12282842546701431, + "grad_norm": 0.07463426887989044, + "kept_token_count": 654.5703125, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9881318807601929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011868099309504032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.0637481689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.44146728515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11790, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20022838562726974, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.12821293249726295, + "grad_norm": 0.06178250536322594, + "kept_token_count": 654.575, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883660793304443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01163390902802348, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.36884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8001434326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11800, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19977982938289643, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.12991279289126395, + "grad_norm": 0.06642169505357742, + "kept_token_count": 654.825, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885405540466309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011459455452859401, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.4803497314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.75446166992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11810, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.1999034985899925, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.13994516655802727, + "grad_norm": 0.0722138062119484, + "kept_token_count": 654.60625, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878324091434478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012167590856552123, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.0746215820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3397491455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11820, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19985502511262893, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.12343579679727554, + "grad_norm": 0.06434379518032074, + "kept_token_count": 654.8546875, + "learning_rate": 2.128327444136365e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9870062291622161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012993785738945007, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.2946258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3001739501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11830, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.1992720529437065, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.12704130113124848, + "grad_norm": 0.08784549683332443, + "kept_token_count": 655.028125, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880827128887176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011917284317314624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.83648376464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.87962341308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11840, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20018159449100495, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.12463062852621079, + "grad_norm": 0.07415053248405457, + "kept_token_count": 654.4796875, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878313839435577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012168618384748697, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.9146453857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.52557983398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11850, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19969556331634522, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.12127480283379555, + "grad_norm": 0.06426308304071426, + "kept_token_count": 654.9203125, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886134326457977, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01138656185939908, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.4495391845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.21788330078124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11860, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1997743770480156, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.12069148644804954, + "grad_norm": 0.053096022456884384, + "kept_token_count": 655.05, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888122081756592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011187791731208563, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.75059814453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.29404602050784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11870, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.2003838911652565, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.12941398546099664, + "grad_norm": 0.06037573516368866, + "kept_token_count": 654.271875, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.989066618680954, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010933385509997607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.44964599609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2687469482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11880, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.19994480162858963, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.12924908399581908, + "grad_norm": 0.04899587482213974, + "kept_token_count": 654.7640625, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883809745311737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01161903776228428, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.773583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9983795166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11890, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19977314323186873, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.12703774347901345, + "grad_norm": 0.08571171015501022, + "kept_token_count": 654.9171875, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.988380491733551, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01161950072273612, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.9200134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3257568359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11900, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.19949525743722915, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.11872471496462822, + "grad_norm": 0.06959404796361923, + "kept_token_count": 654.99375, + "learning_rate": 2.116291843968583e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9879882991313934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012011700496077537, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.95386962890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.42645263671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11910, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.1998648464679718, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.13136940747499465, + "grad_norm": 0.06891706585884094, + "kept_token_count": 654.7890625, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882104635238648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01178952744230628, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.1968658447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70201416015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11920, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20024017691612245, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.13413472920656205, + "grad_norm": 0.0479414239525795, + "kept_token_count": 654.178125, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9891497194766998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010850274097174406, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.73851013183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2416229248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11930, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.20042974203824998, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.12688078135251998, + "grad_norm": 0.0571267195045948, + "kept_token_count": 654.265625, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882367730140686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011763216648250818, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.112548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.74005432128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11940, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19959452748298645, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.125200504809618, + "grad_norm": 0.06401766091585159, + "kept_token_count": 655.0484375, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886192142963409, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011380772851407529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.08042907714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.916748046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.20004227608442307, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.13253100216388702, + "grad_norm": 0.07376197725534439, + "kept_token_count": 654.4734375, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882221996784211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011777795385569334, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.77820739746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.91266479492185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11960, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2005309835076332, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.12519798576831817, + "grad_norm": 0.07050859183073044, + "kept_token_count": 654.3265625, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882489800453186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011751020699739457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.18878173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.11753845214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11970, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.19955519884824752, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.12803720459342002, + "grad_norm": 0.05516904219985008, + "kept_token_count": 655.240625, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882723689079285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011727622523903847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.24566650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.54246215820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11980, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.20020409673452377, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.11552498117089272, + "grad_norm": 0.05365615710616112, + "kept_token_count": 654.3015625, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887601733207703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011239834409207106, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.4221557617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.09385986328124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11990, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19988541156053544, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.12866943553090096, + "grad_norm": 0.049929168075323105, + "kept_token_count": 654.821875, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889827132225036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011017273273319006, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.96884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0509552001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12000, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20000177174806594, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.13585799410939217, + "grad_norm": 0.04453754052519798, + "kept_token_count": 654.5171875, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885412156581879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011458787973970175, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.0263641357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4816162109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12010, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.1999130979180336, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.11608513817191124, + "grad_norm": 0.08864954859018326, + "kept_token_count": 654.6828125, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883443117141724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011655705142766237, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.1783905029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2612243652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12020, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.20011270940303802, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.12199297919869423, + "grad_norm": 0.05268023535609245, + "kept_token_count": 654.2328125, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884022533893585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011597751546651125, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.570947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3503936767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12030, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.2001139149069786, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.12302876785397529, + "grad_norm": 0.04907127469778061, + "kept_token_count": 654.4390625, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893370270729065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010662980284541846, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.9334197998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1994689941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20034164190292358, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.12279043644666672, + "grad_norm": 0.05854036659002304, + "kept_token_count": 654.4421875, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889003157615661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011099684797227383, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1757080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.50534973144534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20047426223754883, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.1312938906252384, + "grad_norm": 0.06889360398054123, + "kept_token_count": 654.3734375, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883056521415711, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011694353632628917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0025177001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9009246826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12060, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.2002839371562004, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.12977738082408904, + "grad_norm": 0.0863930881023407, + "kept_token_count": 654.2328125, + "learning_rate": 2.092077804513201e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889500379562378, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011049942392855883, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.1049468994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9920318603516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12070, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20041164457798005, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.11645994260907173, + "grad_norm": 0.0687653198838234, + "kept_token_count": 654.4609375, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9876600563526153, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01233995808288455, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.08062133789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81085510253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12080, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.19985590875148773, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.12556299343705177, + "grad_norm": 0.049912918359041214, + "kept_token_count": 654.784375, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885583996772767, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011441594082862138, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.0676239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.02709045410154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12090, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20032721906900405, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.13135019987821578, + "grad_norm": 0.05007505789399147, + "kept_token_count": 654.3703125, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882897317409516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011710263136774302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.72100524902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3302795410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12100, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19978051036596298, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.13150802180171012, + "grad_norm": 0.039114173501729965, + "kept_token_count": 654.525, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888201117515564, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011179881636053324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.7461242675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2384735107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12110, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.20047181099653244, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.11147836372256278, + "grad_norm": 0.05877246335148811, + "kept_token_count": 654.4046875, + "learning_rate": 2.084472792508167e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889831662178039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011016849614679813, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.2058074951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12120, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.1998435914516449, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.12880130633711814, + "grad_norm": 0.05501184239983559, + "kept_token_count": 654.8765625, + "learning_rate": 2.082949652141508e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880192101001739, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011980788223445416, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.27672424316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.19130554199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12130, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20094679892063141, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.11635831147432327, + "grad_norm": 0.056353792548179626, + "kept_token_count": 653.8859375, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9877599954605103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012239991221576929, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9415222167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.64295349121096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12140, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19914332330226897, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.12008891105651856, + "grad_norm": 0.05074017122387886, + "kept_token_count": 655.3234375, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894341707229615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01056582322344184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.2246948242188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1955047607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12150, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.20014028549194335, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.12632423937320708, + "grad_norm": 0.06984768807888031, + "kept_token_count": 654.3859375, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891752123832702, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01082480289041996, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.40403137207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3859832763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12160, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20002676844596862, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.12331935614347458, + "grad_norm": 0.08187852799892426, + "kept_token_count": 654.3875, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884014666080475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011598521377891302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.53676452636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.15647888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12170, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.20034165680408478, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.13082863241434098, + "grad_norm": 0.08571894466876984, + "kept_token_count": 654.40625, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.989067530632019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01093246303498745, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.917529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.13181457519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12180, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19910873025655745, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.11923735737800598, + "grad_norm": 0.051432471722364426, + "kept_token_count": 655.4375, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0121, + "mgd_cos_sim": 0.9878836810588837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012116319220513105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.095947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.64440002441404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12190, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.2001497268676758, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.12551698163151742, + "grad_norm": 0.06689983606338501, + "kept_token_count": 654.425, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886398017406464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011360199190676212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9496551513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5426025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12200, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19934405535459518, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.13221389278769494, + "grad_norm": 0.07222741097211838, + "kept_token_count": 655.09375, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889490187168122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011050983052700757, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.600537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83746948242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12210, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19992135018110274, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.12326818257570267, + "grad_norm": 0.0588064081966877, + "kept_token_count": 654.690625, + "learning_rate": 2.069209735969855e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9890123248100281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010987668391317129, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.7056427001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.90765991210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12220, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19926711767911912, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.1228676050901413, + "grad_norm": 0.04918813705444336, + "kept_token_count": 655.184375, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.98838672041893, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01161327762529254, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.12676696777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.08270568847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12230, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1998796731233597, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.1253478743135929, + "grad_norm": 0.05789664387702942, + "kept_token_count": 654.6078125, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886651754379272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011334838252514601, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.98631896972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8609161376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12240, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.20065305680036544, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.12186488062143326, + "grad_norm": 0.05548988655209541, + "kept_token_count": 654.1203125, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890587449073791, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010941268503665924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.1623901367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7301330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19976604878902435, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.11752542778849602, + "grad_norm": 0.08634628355503082, + "kept_token_count": 654.83125, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9874298751354218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012570120114833116, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.6381805419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4093414306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12260, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19987602829933165, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.13428751826286317, + "grad_norm": 0.0442122183740139, + "kept_token_count": 654.9984375, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883742511272431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01162575101479888, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4727386474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.13099670410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12270, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.20013150572776794, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.12304272130131721, + "grad_norm": 0.05669068172574043, + "kept_token_count": 654.828125, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9882169365882874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011783048138022422, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0855438232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.14458618164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12280, + "valid_token_count": 818.6703125 + }, + { + "actual_token_mask_ratio": 0.1996803879737854, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.12120754718780517, + "grad_norm": 0.06118335947394371, + "kept_token_count": 655.1078125, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888493657112122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011150650680065155, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.7752746582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.79534606933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12290, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.20055708885192872, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.13142325058579446, + "grad_norm": 0.05558430776000023, + "kept_token_count": 654.378125, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892875850200653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010712412372231483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.22779846191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.86488647460936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12300, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2001856729388237, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.12556527107954024, + "grad_norm": 0.0595727264881134, + "kept_token_count": 654.4515625, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884213924407959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011578609701246023, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.88149108886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0248565673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12310, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.1991267755627632, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.1250796541571617, + "grad_norm": 0.06277590245008469, + "kept_token_count": 655.1046875, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882544159889222, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011745581962168216, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.25706481933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9891876220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12320, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.2002582997083664, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.11777347102761268, + "grad_norm": 0.07406783103942871, + "kept_token_count": 654.703125, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884139716625213, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01158602787181735, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.0178955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66590881347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12330, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.20061076879501344, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.11983660161495209, + "grad_norm": 0.057872556149959564, + "kept_token_count": 653.9953125, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887048065662384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011295205540955066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.5209075927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0283721923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12340, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19989010989665984, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.13256317973136902, + "grad_norm": 0.07772199809551239, + "kept_token_count": 654.65625, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888059079647065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011194096319377423, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.25648193359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.28802490234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.19976803213357924, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.12584985494613649, + "grad_norm": 0.05473272129893303, + "kept_token_count": 655.1265625, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885636627674103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01143635381013155, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.23875732421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0447601318359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12360, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.200395467877388, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.11930945068597794, + "grad_norm": 0.047875046730041504, + "kept_token_count": 654.3609375, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.988095635175705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011904364079236984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.79417724609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8419921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12370, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19981688559055327, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.11797462403774261, + "grad_norm": 0.06484673172235489, + "kept_token_count": 655.0015625, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889751315116883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011024870350956917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.7222137451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3080810546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12380, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20044079571962356, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.11678210794925689, + "grad_norm": 0.05773799866437912, + "kept_token_count": 654.253125, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885266959667206, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011473307013511657, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2433624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4105743408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12390, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.19943679869174957, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.12281041666865349, + "grad_norm": 0.09308316558599472, + "kept_token_count": 655.4171875, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892573475837707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010742662008851767, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.6862426757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.96904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12400, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.19961692094802858, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.12104131057858467, + "grad_norm": 0.05493234097957611, + "kept_token_count": 655.2890625, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889108419418335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011089148558676242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.58607482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.46168518066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12410, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.19991747587919234, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.11656490713357925, + "grad_norm": 0.055394090712070465, + "kept_token_count": 654.575, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895677208900452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010432290751487017, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.29847717285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4687255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12420, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2008360281586647, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.13573764339089395, + "grad_norm": 0.05218175798654556, + "kept_token_count": 653.7796875, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885313808917999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011468617524951697, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7382568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6955963134766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12430, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20098621100187303, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.11352692097425461, + "grad_norm": 0.057177577167749405, + "kept_token_count": 654.1015625, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889408230781556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011059189680963755, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.3030944824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.31578063964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12440, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.2005677491426468, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.12973795905709268, + "grad_norm": 0.07687294483184814, + "kept_token_count": 654.2921875, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9884866952896119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011513313557952643, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0268829345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3608062744141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12450, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.2000363513827324, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.11914834827184677, + "grad_norm": 0.055155228823423386, + "kept_token_count": 654.7375, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897270739078522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010272931307554245, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.5794281005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6056365966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12460, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.1995575398206711, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.12922822758555413, + "grad_norm": 0.061256665736436844, + "kept_token_count": 655.1015625, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893271923065186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010672780964523555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.82778015136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.45436706542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12470, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.2004975661635399, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.1257053606212139, + "grad_norm": 0.053536735475063324, + "kept_token_count": 654.0453125, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892980873584747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010701915808022022, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.5187225341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8965270996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12480, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20018564015626908, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.1235855646431446, + "grad_norm": 0.05931058153510094, + "kept_token_count": 654.5046875, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893588483333587, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010641139186918735, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.8017150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2716033935547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12490, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19996913224458696, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.12326738983392715, + "grad_norm": 0.06318126618862152, + "kept_token_count": 654.5734375, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884146690368653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011585334315896034, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.22992248535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.22484436035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12500, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19939142763614653, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.12307122349739075, + "grad_norm": 0.04827294871211052, + "kept_token_count": 655.3453125, + "learning_rate": 2.024562430423576e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889710068702697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01102897459641099, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.3783355712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2008819580078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12510, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.1997921124100685, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.11916969791054725, + "grad_norm": 0.09056815505027771, + "kept_token_count": 654.71875, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.988671624660492, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01132837189361453, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.1118865966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1487243652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12520, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20051200240850447, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.12601403146982193, + "grad_norm": 0.054615408182144165, + "kept_token_count": 654.3140625, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885236859321594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01147630987688899, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.63953552246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9497375488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12530, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2002084195613861, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.1263485498726368, + "grad_norm": 0.062459759414196014, + "kept_token_count": 654.4421875, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884010672569274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01159894121810794, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.48402709960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8644958496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2005678340792656, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.11938741728663445, + "grad_norm": 0.09565503150224686, + "kept_token_count": 653.9875, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884005725383759, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011599416099488735, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.41588134765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.52557678222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12550, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.20011191070079803, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.12320334389805794, + "grad_norm": 0.06062386557459831, + "kept_token_count": 654.34375, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.989296019077301, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01070398036390543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.35459594726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5911376953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12560, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.19912571012973784, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.12835070937871934, + "grad_norm": 0.04774833098053932, + "kept_token_count": 655.328125, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887688279151916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01123115113005042, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8054138183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.71905822753905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12570, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2005230262875557, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.13580354824662208, + "grad_norm": 0.05990985408425331, + "kept_token_count": 654.2578125, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897305727005005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010269421339035033, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.7661834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0557800292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12580, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20179806649684906, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.12952580526471139, + "grad_norm": 0.0601000040769577, + "kept_token_count": 653.2046875, + "learning_rate": 2.012149654706343e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889541208744049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011045871395617723, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.24823608398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.33783264160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20085201263427735, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.12179943397641183, + "grad_norm": 0.06596924364566803, + "kept_token_count": 653.98125, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893608212471008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010639180988073349, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.87343139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1219909667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12600, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20049380958080293, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.12927970588207244, + "grad_norm": 0.061338867992162704, + "kept_token_count": 654.134375, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889570891857147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011042924784123897, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.89940490722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.89398498535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12610, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19976455867290496, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.12642953395843506, + "grad_norm": 0.08220665156841278, + "kept_token_count": 654.7703125, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886208951473237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011379101406782865, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.06638488769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.52369995117186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12620, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.19969508647918702, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.1289665624499321, + "grad_norm": 0.08706815540790558, + "kept_token_count": 654.6890625, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893140375614167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010685952845960855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.0935455322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.86207275390626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.20049229860305787, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.13263634517788886, + "grad_norm": 0.06808600574731827, + "kept_token_count": 654.0546875, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.011, + "mgd_cos_sim": 0.989025992155075, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010974001418799163, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.6322784423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7080139160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12640, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.2001192107796669, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.11344664245843887, + "grad_norm": 0.07068468630313873, + "kept_token_count": 654.375, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891572237014771, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010842785984277726, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.74013366699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.60735473632815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12650, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19993281066417695, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.12929273322224616, + "grad_norm": 0.0607646182179451, + "kept_token_count": 654.66875, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.989673274755478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010326720587909222, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.58214111328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5052764892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19987493008375168, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.11946560069918633, + "grad_norm": 0.04022250697016716, + "kept_token_count": 654.6046875, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890949547290802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010905032232403754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.52155151367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7013732910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12670, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2002589538693428, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.12001289054751396, + "grad_norm": 0.06197621300816536, + "kept_token_count": 654.7140625, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893124103546143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010687588527798653, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.6609252929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.10834350585935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12680, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.19993706196546554, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.1240301288664341, + "grad_norm": 0.05058739706873894, + "kept_token_count": 654.9609375, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.989595377445221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010404601134359837, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.27674560546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5533782958984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12690, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.19969775378704072, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.11385648101568221, + "grad_norm": 0.04443764314055443, + "kept_token_count": 654.771875, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897499442100525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010250054765492678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.8529937744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7357116699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12700, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20100192129611968, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.12664092034101487, + "grad_norm": 0.05550149455666542, + "kept_token_count": 653.9, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891717374324799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010828257631510496, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.38657531738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07245178222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12710, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2002138689160347, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.12717511355876923, + "grad_norm": 0.07251759618520737, + "kept_token_count": 654.2796875, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.989127004146576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01087299846112728, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.8109619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0583099365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12720, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.20105232298374176, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.12701538503170012, + "grad_norm": 0.058485761284828186, + "kept_token_count": 653.6265625, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886562526226044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011343746911734343, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.38844909667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1274719238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12730, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20047890841960908, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.1206948310136795, + "grad_norm": 0.04832396283745766, + "kept_token_count": 654.2390625, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895522356033325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010447781160473823, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.59963073730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8475677490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12740, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20075605809688568, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.13019668012857438, + "grad_norm": 0.0728103518486023, + "kept_token_count": 653.85, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895829558372498, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01041703624650836, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.4367218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55933837890626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12750, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.20083339512348175, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.1282892867922783, + "grad_norm": 0.06491892039775848, + "kept_token_count": 654.0328125, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885159254074096, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01148406956344843, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.2601623535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0710388183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12760, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2000437632203102, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.13009707927703856, + "grad_norm": 0.06167469918727875, + "kept_token_count": 654.5515625, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889898002147675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0110102036036551, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.21178283691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1516967773438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12770, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2003485843539238, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.1240907333791256, + "grad_norm": 0.08231153339147568, + "kept_token_count": 654.275, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889801919460297, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011019802279770374, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.79486389160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3160034179688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12780, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.1997077226638794, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.13977532759308814, + "grad_norm": 0.10312171280384064, + "kept_token_count": 654.703125, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9877147316932678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012285269610583783, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.1549865722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.08692321777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12790, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.19973253458738327, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.12343726307153702, + "grad_norm": 0.06472531706094742, + "kept_token_count": 654.565625, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891736268997192, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010826377011835576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8179016113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3607604980469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12800, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.19962515234947203, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.11856156066060067, + "grad_norm": 0.0511530302464962, + "kept_token_count": 654.959375, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888082563877105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011191741097718477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.66236572265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5925354003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12810, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19994690120220185, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.12540634572505951, + "grad_norm": 0.0542265959084034, + "kept_token_count": 654.8796875, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890858292579651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010914173629134893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.1563995361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6625091552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12820, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19967144280672072, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.12587133944034576, + "grad_norm": 0.07001641392707825, + "kept_token_count": 655.2, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.988432776927948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01156722493469715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.74269714355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8753723144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12830, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.19999036341905593, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.12168865427374839, + "grad_norm": 0.054803717881441116, + "kept_token_count": 654.746875, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0122, + "mgd_cos_sim": 0.9878416657447815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.012158333230763674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.7992919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.42548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12840, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.19947845339775086, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.13128421157598497, + "grad_norm": 0.06165311112999916, + "kept_token_count": 655.190625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9883424460887908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011657565087080001, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.764111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.07843017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12850, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.19960186183452605, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.12116971015930175, + "grad_norm": 0.0599200539290905, + "kept_token_count": 655.0171875, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887667894363403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01123321456834674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.5924468994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.95791015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12860, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19973719269037246, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.125402595102787, + "grad_norm": 0.057283930480480194, + "kept_token_count": 654.725, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887630879878998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011236912105232477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.87437744140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.06504821777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12870, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.199774631857872, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.12631720975041388, + "grad_norm": 0.05093997344374657, + "kept_token_count": 654.6234375, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888432443141937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011156752146780491, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.55437622070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6675231933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12880, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20007171034812926, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.13622836098074914, + "grad_norm": 0.09087395668029785, + "kept_token_count": 654.4265625, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892695307731628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0107304728589952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.9351470947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.37159118652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12890, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20006952285766602, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.12573407888412474, + "grad_norm": 0.052992962300777435, + "kept_token_count": 654.5453125, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894967555999756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010503249522298575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.7820587158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2963592529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.19960614740848542, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.11379125267267227, + "grad_norm": 0.06172899529337883, + "kept_token_count": 654.9578125, + "learning_rate": 1.962109974767429e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9890174627304077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010982533730566502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.70110473632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30368957519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.19979888796806336, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.12548799961805343, + "grad_norm": 0.05353418365120888, + "kept_token_count": 654.7484375, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898585677146912, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010141425672918558, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4125183105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7688720703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12920, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.20057819336652755, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.13072569370269777, + "grad_norm": 0.05863667279481888, + "kept_token_count": 654.24375, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0118, + "mgd_cos_sim": 0.9881975471973419, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011802462209016085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.49135437011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0286437988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12930, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20002305656671523, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.13368804901838302, + "grad_norm": 0.06126817315816879, + "kept_token_count": 654.7140625, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890970289707184, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010902974847704172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.15789184570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.04127197265626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12940, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.1996353343129158, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.12281139567494392, + "grad_norm": 0.06697249412536621, + "kept_token_count": 655.15, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886616230010986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011338370200246573, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.9219635009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5733245849609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12950, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.19981313347816468, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.12181826159358025, + "grad_norm": 0.05724642798304558, + "kept_token_count": 654.7234375, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.988379979133606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011620016861706972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.7157012939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6451171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12960, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19994160085916518, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.11736488044261932, + "grad_norm": 0.04036755487322807, + "kept_token_count": 654.4359375, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894015848636627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010598415974527597, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.0609619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.37932739257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12970, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.19969086349010468, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.12009403184056282, + "grad_norm": 0.0651477724313736, + "kept_token_count": 654.9734375, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894526362419128, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010547368787229061, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.8137145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3226684570312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12980, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19993204921483992, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.12167729362845421, + "grad_norm": 0.05431775003671646, + "kept_token_count": 654.7078125, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9892148673534393, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010785123892128467, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.41510009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6369323730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19957152307033538, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.1254281558096409, + "grad_norm": 0.06314387172460556, + "kept_token_count": 654.903125, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.012, + "mgd_cos_sim": 0.9880151450634003, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011984848976135254, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.478076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.92250671386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13000, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19951253086328508, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.12326994463801384, + "grad_norm": 0.12512436509132385, + "kept_token_count": 655.1046875, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887338578701019, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011266139335930347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.815576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.96869201660155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13010, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20016818195581437, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.12265088185667991, + "grad_norm": 0.06220122426748276, + "kept_token_count": 654.58125, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0115, + "mgd_cos_sim": 0.9885263681411743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011473626550287008, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.84625549316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.73517150878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13020, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20070562660694122, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.12276658490300178, + "grad_norm": 0.058631353080272675, + "kept_token_count": 654.0328125, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9892291784286499, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010770828742533923, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.22984924316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.99656372070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13030, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19967761933803557, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.12073353677988052, + "grad_norm": 0.04933803901076317, + "kept_token_count": 654.784375, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886296689510345, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011370335426181554, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.66188659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1407470703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13040, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20032331049442292, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.12630084827542304, + "grad_norm": 0.06564606726169586, + "kept_token_count": 654.4328125, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895589470863342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01044105114415288, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.90240173339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0898773193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13050, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.20019926279783248, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.12806011810898782, + "grad_norm": 0.06709811836481094, + "kept_token_count": 654.415625, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898692965507507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010130705684423447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9455596923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5152069091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13060, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.1993905484676361, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.1195767104625702, + "grad_norm": 0.05049695074558258, + "kept_token_count": 655.325, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888616323471069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011138370074331761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.9886810302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1835906982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13070, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.199267153441906, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.1292945474386215, + "grad_norm": 0.055061522871255875, + "kept_token_count": 655.2703125, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887514531612396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011248551215976477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.35613708496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.30787963867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13080, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.20116236805915833, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.13109009563922883, + "grad_norm": 0.06712295114994049, + "kept_token_count": 654.0625, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896835625171662, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01031643282622099, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0402038574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8068817138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13090, + "valid_token_count": 818.7703125 + }, + { + "actual_token_mask_ratio": 0.200307297706604, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.12049056068062783, + "grad_norm": 0.07561037689447403, + "kept_token_count": 654.3125, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895890355110168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010410964954644442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.94029235839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9624359130859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13100, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.19916597604751587, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.134578076004982, + "grad_norm": 0.05756643787026405, + "kept_token_count": 655.171875, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888243198394775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01117568202316761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.3811431884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0233581542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13110, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.20009433776140212, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.13035853430628777, + "grad_norm": 0.06432411074638367, + "kept_token_count": 654.6546875, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897098422050477, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010290147457271814, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.95054016113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.130322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13120, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.1997002437710762, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.12488805055618286, + "grad_norm": 0.06065122038125992, + "kept_token_count": 654.9703125, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886631011962891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01133690308779478, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3346374511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.61659240722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13130, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2002147227525711, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.11731980517506599, + "grad_norm": 0.08747708797454834, + "kept_token_count": 654.5046875, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.989574772119522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010425216052681208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.38816528320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6808135986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.19993155747652053, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.12227005213499069, + "grad_norm": 0.05990613251924515, + "kept_token_count": 654.6328125, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887156546115875, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011284325178712607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.7588195800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4050720214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13150, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2000308081507683, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.1301768623292446, + "grad_norm": 0.07088453322649002, + "kept_token_count": 654.740625, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890836358070374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010916364192962647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.4398986816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.62037963867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13160, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.19874792844057082, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.1331166423857212, + "grad_norm": 0.056767091155052185, + "kept_token_count": 655.7578125, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886248230934143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011375159583985805, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.2451690673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5111022949219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13170, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20016237944364548, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.1252104125916958, + "grad_norm": 0.07590765506029129, + "kept_token_count": 654.3359375, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888648509979248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01113515580072999, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.2860992431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3651458740234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13180, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19917895197868346, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.12834135964512824, + "grad_norm": 0.08192232251167297, + "kept_token_count": 655.184375, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0119, + "mgd_cos_sim": 0.9880674660205842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011932540126144886, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.6011566162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9867462158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13190, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.1993722438812256, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.1197175294160843, + "grad_norm": 0.06056736037135124, + "kept_token_count": 654.9125, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9891498506069183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010850140172988177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.2733551025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2306732177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13200, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.1998405337333679, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.12496638596057892, + "grad_norm": 0.038246870040893555, + "kept_token_count": 654.7296875, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894593775272369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010540629271417856, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.39727172851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8702880859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13210, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20008181631565095, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.12304896935820579, + "grad_norm": 0.057142604142427444, + "kept_token_count": 654.2109375, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891929149627685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010807066317647695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.77496948242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.94952392578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13220, + "valid_token_count": 817.846875 + }, + { + "actual_token_mask_ratio": 0.19889231622219086, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.12724918872117996, + "grad_norm": 0.046365611255168915, + "kept_token_count": 655.5796875, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897421360015869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010257871262729168, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.05787658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.39215393066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13230, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20107713639736174, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.12767402082681656, + "grad_norm": 0.04681834951043129, + "kept_token_count": 653.5625, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.989533132314682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010466869082301855, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9881317138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3919952392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13240, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.1997154399752617, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.11778577789664268, + "grad_norm": 0.055295221507549286, + "kept_token_count": 654.8703125, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891857326030731, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010814265348017216, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.33469848632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3154357910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13250, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.19965148717164993, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.12081804722547532, + "grad_norm": 0.05450265482068062, + "kept_token_count": 655.1828125, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898220002651215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010178015101701021, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.9079895019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.39524841308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13260, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.19998400360345842, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.12184074446558953, + "grad_norm": 0.05054766684770584, + "kept_token_count": 654.1109375, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895281553268432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010471823811531066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.29652404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7523986816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13270, + "valid_token_count": 817.621875 + }, + { + "actual_token_mask_ratio": 0.20036267936229707, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.12250434830784798, + "grad_norm": 0.05623699352145195, + "kept_token_count": 654.525, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900462388992309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00995376417413354, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.09203186035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.38758544921876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13280, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.20062865167856217, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.12379397973418235, + "grad_norm": 0.06550364196300507, + "kept_token_count": 654.053125, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887157797813415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011284222267568112, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.3653625488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7999542236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13290, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.1992355540394783, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.12548650577664375, + "grad_norm": 0.055034760385751724, + "kept_token_count": 655.1046875, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989885127544403, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010114871710538865, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.20838928222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13300, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20094375908374787, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.12876226305961608, + "grad_norm": 0.05856034532189369, + "kept_token_count": 653.671875, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0117, + "mgd_cos_sim": 0.9882966578006744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011703351326286793, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6844024658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.77998046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13310, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2001548781991005, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.1388073518872261, + "grad_norm": 0.05799243971705437, + "kept_token_count": 654.690625, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.98946812748909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010531868785619736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.65776977539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7276123046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13320, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.19992603957653046, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.12602048739790916, + "grad_norm": 0.06516663730144501, + "kept_token_count": 654.6890625, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889912068843841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011008796468377113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.58739318847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9665466308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13330, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19870675802230836, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.11855846717953682, + "grad_norm": 0.05792788043618202, + "kept_token_count": 655.5875, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894298911094666, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010570120066404343, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.39813232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63653869628905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13340, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20053300112485886, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.12102960050106049, + "grad_norm": 0.07254160195589066, + "kept_token_count": 653.8109375, + "learning_rate": 1.892379503427575e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900469660758973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00995303075760603, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.83490295410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.71143493652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13350, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.19926805049180984, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.1303282156586647, + "grad_norm": 0.06775948405265808, + "kept_token_count": 655.31875, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894176721572876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010582329239696264, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.6771514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6914520263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13360, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.20038514733314514, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.10754309371113777, + "grad_norm": 0.06693115085363388, + "kept_token_count": 654.296875, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904875695705414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00951244281604886, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.8319152832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.43389892578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20055551677942277, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.12763548716902734, + "grad_norm": 0.053926050662994385, + "kept_token_count": 654.559375, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901230156421661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00987698184326291, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.08409423828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1282623291016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13380, + "valid_token_count": 818.7625 + }, + { + "actual_token_mask_ratio": 0.1998043179512024, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.1155525952577591, + "grad_norm": 0.08291811496019363, + "kept_token_count": 654.925, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893651902675629, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010634817276149988, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.8915771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3730773925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.20021575540304185, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.12493751049041749, + "grad_norm": 0.06180022284388542, + "kept_token_count": 654.41875, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891839146614074, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010816088411957026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0506896972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5642852783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13400, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.2004888251423836, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.1231192633509636, + "grad_norm": 0.06852652132511139, + "kept_token_count": 654.3484375, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9883811712265015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011618824023753405, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.84228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7002807617188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13410, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20010766088962556, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.13183390349149704, + "grad_norm": 0.06402057409286499, + "kept_token_count": 654.6390625, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889678955078125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011032102908939124, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.7549560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7152404785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13420, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20004975646734238, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.12841543778777123, + "grad_norm": 0.048627302050590515, + "kept_token_count": 654.83125, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900801122188568, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00991988405585289, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.57915649414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8106414794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13430, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.199870964884758, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.13002714216709138, + "grad_norm": 0.057320717722177505, + "kept_token_count": 654.753125, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896596610546112, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010340321343392134, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.5377685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31663208007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13440, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20054442882537843, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.12481139972805977, + "grad_norm": 0.059212226420640945, + "kept_token_count": 654.196875, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893644273281097, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010635594837367535, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5397644042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5725341796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13450, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.1995772749185562, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.12129504904150963, + "grad_norm": 0.057981252670288086, + "kept_token_count": 654.8078125, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901272475719451, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009872758947312831, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0359680175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98399658203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13460, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19980425834655763, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.12365960627794266, + "grad_norm": 0.07217594981193542, + "kept_token_count": 654.740625, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0116, + "mgd_cos_sim": 0.9884322166442872, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011567782703787088, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.83135375976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2694915771484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13470, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19947753697633744, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.12068402916193008, + "grad_norm": 0.0486648790538311, + "kept_token_count": 655.103125, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895458161830902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01045417357236147, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9228576660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8778778076172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19863308519124984, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.1262631081044674, + "grad_norm": 0.04990638047456741, + "kept_token_count": 655.6484375, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895230531692505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010476933605968953, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.2237609863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.37653198242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19939233958721161, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.12311701849102974, + "grad_norm": 0.05301530286669731, + "kept_token_count": 654.9984375, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989859277009964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010140712652355433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.6707702636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.88511047363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13500, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20009214282035828, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.13235804736614226, + "grad_norm": 0.0692065879702568, + "kept_token_count": 654.3921875, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887768030166626, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011223199125379325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7251708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.979296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13510, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.19979507625102996, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.12009242922067642, + "grad_norm": 0.0960124284029007, + "kept_token_count": 654.7265625, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893493473529815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010650651156902313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.18796691894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13520, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.1994418516755104, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.12186802104115486, + "grad_norm": 0.05508685111999512, + "kept_token_count": 655.0703125, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890948414802552, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010905170906335115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.4244873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2969085693359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20040500313043594, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.122274848818779, + "grad_norm": 0.09368672966957092, + "kept_token_count": 654.0265625, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898246586322784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010175363346934318, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.6669097900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0297393798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13540, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.19979803413152694, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.12306059673428535, + "grad_norm": 0.06167398393154144, + "kept_token_count": 654.84375, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897550046443939, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010244990419596433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.32318420410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7512939453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13550, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.200052846968174, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.12968803569674492, + "grad_norm": 0.05938296765089035, + "kept_token_count": 654.6375, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889278173446655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011072190199047326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.33156127929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.06102294921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13560, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19880366772413255, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.12518999949097634, + "grad_norm": 0.0567685067653656, + "kept_token_count": 655.653125, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9892458379268646, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010754159558564425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.0196533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8234130859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13570, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2002081573009491, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.13001147732138635, + "grad_norm": 0.05610829219222069, + "kept_token_count": 654.4875, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892895460128784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010710457991808652, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.44141845703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2733215332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13580, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.20001427978277206, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.12531251907348634, + "grad_norm": 0.05776253342628479, + "kept_token_count": 654.421875, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.989055198431015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010944796819239854, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5223693847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2250579833984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13590, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.19953292608261108, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.12093558833003044, + "grad_norm": 0.06553332507610321, + "kept_token_count": 654.8578125, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895608067512512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0104391910135746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.7003509521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.26658325195314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13600, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20042076855897903, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.11765437126159668, + "grad_norm": 0.05333859100937843, + "kept_token_count": 654.4015625, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887402176856994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011259780637919903, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.39307556152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2189025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13610, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.1985141158103943, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.11672220081090927, + "grad_norm": 0.055323123931884766, + "kept_token_count": 655.6765625, + "learning_rate": 1.849122896184354e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889848232269287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011015173234045505, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.63861083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3196258544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13620, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19967816323041915, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.11909876689314843, + "grad_norm": 0.07384985685348511, + "kept_token_count": 655.1171875, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896605312824249, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010339472908526659, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.13397216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.90643310546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13630, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20017971098423004, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.11890256330370903, + "grad_norm": 0.05481597036123276, + "kept_token_count": 654.6046875, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900069236755371, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00999309252947569, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.8779052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9738311767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13640, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20046510249376298, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.11811580657958984, + "grad_norm": 0.05520745366811752, + "kept_token_count": 654.321875, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891676723957061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01083231745287776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.8320587158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8530975341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13650, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20008146464824678, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.11666513234376907, + "grad_norm": 0.06122971698641777, + "kept_token_count": 654.6046875, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899522244930268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010047785565257072, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.7369689941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.44399108886716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19976876229047774, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.12250375300645829, + "grad_norm": 0.045524198561906815, + "kept_token_count": 655.0703125, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888415098190307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011158499773591757, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8931030273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53753967285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13670, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.20008034110069275, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.13609760701656343, + "grad_norm": 0.05490266531705856, + "kept_token_count": 654.4171875, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9892358899116516, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01076411223039031, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.91941528320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4219512939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2001355081796646, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.12586428821086884, + "grad_norm": 0.05910181626677513, + "kept_token_count": 654.5140625, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9886563777923584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01134360795840621, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.6147399902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.56934814453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13690, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.20003039985895157, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.13155082389712333, + "grad_norm": 0.06582143902778625, + "kept_token_count": 654.496875, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893320977687836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01066789049655199, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.41612243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3483062744141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13700, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.20074924379587172, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.13060857206583024, + "grad_norm": 0.06294405460357666, + "kept_token_count": 653.8390625, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895154595375061, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010484542045742274, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.89976806640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8822845458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.19977725446224212, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.11578061878681183, + "grad_norm": 0.09616199880838394, + "kept_token_count": 655.0921875, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889398872852325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011060126405209304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.4905059814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3728088378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.2001004233956337, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.1186043806374073, + "grad_norm": 0.056651122868061066, + "kept_token_count": 654.2609375, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894206821918488, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010579327121376991, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.87537841796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1567901611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13730, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.20023933947086334, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.11553451716899872, + "grad_norm": 0.04972219094634056, + "kept_token_count": 654.675, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989854347705841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010145652201026678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4609313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.70287170410154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13740, + "valid_token_count": 818.59375 + }, + { + "actual_token_mask_ratio": 0.19927174597978592, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.1308210738003254, + "grad_norm": 0.08003707230091095, + "kept_token_count": 655.2921875, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898644268512726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01013558330014348, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2137115478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4123992919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13750, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19929211884737014, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.13147691935300826, + "grad_norm": 0.06435169279575348, + "kept_token_count": 655.1546875, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9885663866996766, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011433622613549233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.4643157958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0060974121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2001981794834137, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.12012481540441514, + "grad_norm": 0.054470907896757126, + "kept_token_count": 654.865625, + "learning_rate": 1.824955364871982e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900297284126282, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009970273450016975, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.6772216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.33697814941405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13770, + "valid_token_count": 818.7765625 + }, + { + "actual_token_mask_ratio": 0.200668802857399, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.12645064890384675, + "grad_norm": 0.06497931480407715, + "kept_token_count": 654.146875, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896021962165833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01039779782295227, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.2923217773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.09507446289064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.1996031880378723, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.12461964786052704, + "grad_norm": 0.046112142503261566, + "kept_token_count": 654.7265625, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898607432842255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010139246936887503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.37909545898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0926177978516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13790, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2001201555132866, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.12537760958075522, + "grad_norm": 0.05065230652689934, + "kept_token_count": 654.4125, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895636677742005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010436326917260885, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.4950378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.57106628417966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.1995583102107048, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.12201700508594512, + "grad_norm": 0.053809214383363724, + "kept_token_count": 655.1515625, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892978489398956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01070215431973338, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.4978881835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.63890075683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13810, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.20048509836196898, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.12467225193977356, + "grad_norm": 0.06346864998340607, + "kept_token_count": 654.10625, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888829469680787, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011117066070437432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.67274780273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7882049560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13820, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.2006234809756279, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.11915537789463997, + "grad_norm": 0.11081865429878235, + "kept_token_count": 654.1953125, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9888510704040527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011148934997618199, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.7328643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.60093994140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13830, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.200443734228611, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.13542780578136443, + "grad_norm": 0.05821322277188301, + "kept_token_count": 654.3921875, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895714282989502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010428569372743368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0350372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2781982421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13840, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.20011230260133744, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.12414565533399582, + "grad_norm": 0.06582771241664886, + "kept_token_count": 654.35625, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888262808322906, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0111737166531384, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.98662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.65867004394534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13850, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.19940512180328368, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.11082312911748886, + "grad_norm": 0.06227446347475052, + "kept_token_count": 655.2734375, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900656938552856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009934314899146557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.0065063476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.76531677246095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13860, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19946463108062745, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.12497985810041427, + "grad_norm": 0.04571215808391571, + "kept_token_count": 655.1875, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.989440405368805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01055960487574339, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.5088592529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78981323242186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13870, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20027015209197999, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.13244464695453645, + "grad_norm": 0.05009694769978523, + "kept_token_count": 654.5703125, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896366000175476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010363403987139464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0216949462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9796478271484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13880, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.20041603446006775, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.13344535529613494, + "grad_norm": 0.05331004783511162, + "kept_token_count": 653.9875, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893731355667115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010626869834959507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.23243713378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5969635009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13890, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.20017242878675462, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.13027652353048325, + "grad_norm": 0.04633324220776558, + "kept_token_count": 654.4546875, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9888234198093414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011176585499197245, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.96880798339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0745574951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13900, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19874757379293442, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.13308906182646751, + "grad_norm": 0.06315074115991592, + "kept_token_count": 655.4953125, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905047416687012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00949524100869894, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.65284729003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2930145263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20004197508096694, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.12187624126672744, + "grad_norm": 0.05173345282673836, + "kept_token_count": 654.4203125, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895112752914429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01048871772363782, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.0047576904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3694732666016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13920, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.20003620237112046, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.12082792595028877, + "grad_norm": 0.04289187118411064, + "kept_token_count": 654.3953125, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907048046588898, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009295200183987617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.2641571044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.86119384765624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13930, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.20020687729120254, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.11725658550858498, + "grad_norm": 0.056102268397808075, + "kept_token_count": 654.65, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890625834465027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01093740276992321, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.8628173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.46307373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13940, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19974941462278367, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.1275479793548584, + "grad_norm": 0.0610680878162384, + "kept_token_count": 654.696875, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891900897026062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01080991057679057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.930908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6199462890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13950, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19980971068143843, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.13555736392736434, + "grad_norm": 0.050125811249017715, + "kept_token_count": 654.8671875, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903932690620423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009606733266264201, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.13531799316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5157867431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13960, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1991385504603386, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.13088190481066703, + "grad_norm": 0.0746576264500618, + "kept_token_count": 655.09375, + "learning_rate": 1.79259494044161e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899510860443115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010048904176801442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.0363708496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1116577148438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13970, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.20009897500276566, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.13600437268614768, + "grad_norm": 0.05621692165732384, + "kept_token_count": 654.228125, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902330338954926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009766982030123472, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4030029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9504119873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13980, + "valid_token_count": 817.8828125 + }, + { + "actual_token_mask_ratio": 0.2002304181456566, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.11981784179806709, + "grad_norm": 0.043939001858234406, + "kept_token_count": 654.215625, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891629040241241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010837090853601693, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.242626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4790863037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13990, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.2008284017443657, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.11637909561395646, + "grad_norm": 0.0786280706524849, + "kept_token_count": 654.0375, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894259870052338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01057401429861784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.27732543945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9885986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14000, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20039881467819215, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.12972348034381867, + "grad_norm": 0.05027003213763237, + "kept_token_count": 653.84375, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896508812904358, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010349121689796448, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.8680480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6349761962891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14010, + "valid_token_count": 817.7140625 + }, + { + "actual_token_mask_ratio": 0.2003144770860672, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.12084190919995308, + "grad_norm": 0.04253894090652466, + "kept_token_count": 654.3640625, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899143040180206, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01008569048717618, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.81139221191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.24167785644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14020, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19978948533535004, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.12105639576911927, + "grad_norm": 0.037986889481544495, + "kept_token_count": 654.6640625, + "learning_rate": 1.782858365369289e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899702370166779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010029762051999569, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.22495727539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.61802673339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14030, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2002282276749611, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.12042159512639046, + "grad_norm": 0.05251063406467438, + "kept_token_count": 654.4234375, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898259520530701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01017405278980732, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.69215393066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31293029785155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14040, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2009401276707649, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.12769096866250038, + "grad_norm": 0.051374651491642, + "kept_token_count": 653.9640625, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.98981271982193, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010187286976724863, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.28522033691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98939208984376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14050, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.19916409701108934, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.12368938475847244, + "grad_norm": 0.05063584819436073, + "kept_token_count": 655.1828125, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889306604862214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011069342866539955, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4018951416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67243041992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14060, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19975821375846864, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.11538661643862724, + "grad_norm": 0.055004049092531204, + "kept_token_count": 655.2734375, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894201755523682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010579825192689896, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.0203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.24705505371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14070, + "valid_token_count": 818.8453125 + }, + { + "actual_token_mask_ratio": 0.1997015506029129, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.11850975975394248, + "grad_norm": 0.06054031103849411, + "kept_token_count": 654.7640625, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890761911869049, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010923813004046679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.7225738525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4513641357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14080, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.1999444216489792, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.12028307989239692, + "grad_norm": 0.050278764218091965, + "kept_token_count": 654.3515625, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0113, + "mgd_cos_sim": 0.9887061536312103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011293854005634785, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.2476806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2069152832031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14090, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.2007824510335922, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.12093060985207557, + "grad_norm": 0.06366915255784988, + "kept_token_count": 654.0296875, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892796456813813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010720364842563867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.6736358642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6312194824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14100, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.1993579387664795, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.11579681113362313, + "grad_norm": 0.05591846629977226, + "kept_token_count": 654.83125, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896445751190186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010355424135923386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.95277404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.80568542480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14110, + "valid_token_count": 817.8859375 + }, + { + "actual_token_mask_ratio": 0.19949652254581451, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.12261724323034287, + "grad_norm": 0.05286503955721855, + "kept_token_count": 655.08125, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892729580402374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01072704903781414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.94862365722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.38035278320314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14120, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20059439837932586, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.125962495803833, + "grad_norm": 0.05214480683207512, + "kept_token_count": 654.1109375, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897658407688141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010234163980931044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2457641601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49754638671874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14130, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19865255802869797, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.12847595363855363, + "grad_norm": 0.05049324780702591, + "kept_token_count": 655.7453125, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906263709068298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009373641945421696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.85345458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.36611022949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14140, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2006915733218193, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.12383155971765518, + "grad_norm": 0.07133519649505615, + "kept_token_count": 654.34375, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.990079140663147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009920864924788475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9763488769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.91334838867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14150, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.20048988312482835, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.12346081510186195, + "grad_norm": 0.062123969197273254, + "kept_token_count": 654.16875, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0111, + "mgd_cos_sim": 0.9889205694198608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011079440824687481, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.30181579589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.67718811035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14160, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.19986695647239686, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.1254223294556141, + "grad_norm": 0.08613923192024231, + "kept_token_count": 654.79375, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902689099311829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009731083549559117, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.70762634277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05201416015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20044070184230806, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.12151820585131645, + "grad_norm": 0.05025963485240936, + "kept_token_count": 654.2671875, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896654486656189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010334553476423026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.07883605957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.30694580078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14180, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.1993590459227562, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.11942135095596314, + "grad_norm": 0.061297833919525146, + "kept_token_count": 655.109375, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890669405460357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010933068487793208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.8119689941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9363220214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.20046204924583436, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.12278885990381241, + "grad_norm": 0.04910724610090256, + "kept_token_count": 654.178125, + "learning_rate": 1.755205625245195e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900426387786865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009957352839410304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.2011199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.54798278808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20078782588243485, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.11964888647198677, + "grad_norm": 0.06785575300455093, + "kept_token_count": 654.103125, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895884335041046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010411560721695422, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.3262054443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2096466064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14210, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2004824236035347, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.12915381267666817, + "grad_norm": 0.05844202637672424, + "kept_token_count": 654.0015625, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.989748740196228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010251253563910722, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5563201904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.27345275878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14220, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.19980970174074172, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.1351017750799656, + "grad_norm": 0.0905257984995842, + "kept_token_count": 654.8453125, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896470546722412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010352933034300803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.8995422363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.52911376953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14230, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2001853495836258, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.13386727422475814, + "grad_norm": 0.08126262575387955, + "kept_token_count": 654.1859375, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899088203907013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010091185569763184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.46099548339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.50928955078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14240, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.19983671009540557, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.13317371606826783, + "grad_norm": 0.05021992325782776, + "kept_token_count": 654.528125, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893805861473084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010619417950510978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.9909606933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.975390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.1995565950870514, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.12243433892726899, + "grad_norm": 0.06645053625106812, + "kept_token_count": 655.0796875, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894393622875214, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010560640413314105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.3544464111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5375946044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14260, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19949082136154175, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.12617254853248597, + "grad_norm": 0.05607660114765167, + "kept_token_count": 654.9546875, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902263224124909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009773676842451095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.6047637939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8600708007813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14270, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2005808398127556, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.12491915002465248, + "grad_norm": 0.05596606060862541, + "kept_token_count": 654.3421875, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.9887704610824585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011229549720883369, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.7662689208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7226898193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14280, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19993928968906402, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.12663108631968498, + "grad_norm": 0.042419880628585815, + "kept_token_count": 654.7265625, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900093376636505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009990665223449468, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.97210998535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2336059570313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19986737817525863, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.12394513115286827, + "grad_norm": 0.0524996779859066, + "kept_token_count": 654.7234375, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894721806049347, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010527822095900774, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.66136474609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.282568359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14300, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20093071460723877, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.1273372143507004, + "grad_norm": 0.05219843611121178, + "kept_token_count": 653.7359375, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893465757369995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010653423331677913, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.1032348632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.95692443847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14310, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.1999232515692711, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.12385587468743324, + "grad_norm": 0.04341079667210579, + "kept_token_count": 654.846875, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899980366230011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010001949220895767, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.82469787597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1361877441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14320, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2000027373433113, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.12736885100603104, + "grad_norm": 0.07823556661605835, + "kept_token_count": 654.65625, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0114, + "mgd_cos_sim": 0.9886381566524506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011361845582723618, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8390380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6722442626953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14330, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.19950348138809204, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.1382325477898121, + "grad_norm": 0.05862084776163101, + "kept_token_count": 654.725, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901358246803283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009864163119345903, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.54095153808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.32964782714845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14340, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.20062678903341294, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.11823920756578446, + "grad_norm": 0.04483632370829582, + "kept_token_count": 654.0609375, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903654515743255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009634546656161546, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.60337524414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4795196533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20006210952997208, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.12871155962347985, + "grad_norm": 0.062261831015348434, + "kept_token_count": 654.4828125, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894842624664306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010515743214637042, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.7204559326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.97633056640626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14360, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.1989228069782257, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.11842063590884208, + "grad_norm": 0.04943684861063957, + "kept_token_count": 655.8125, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893038749694825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01069612493738532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.03379211425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9347839355469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14370, + "valid_token_count": 818.6671875 + }, + { + "actual_token_mask_ratio": 0.200808584690094, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.1160806767642498, + "grad_norm": 0.050595033913850784, + "kept_token_count": 654.1984375, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901316881179809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009868317283689976, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.6293609619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2650177001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14380, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.20118069797754287, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.12665960416197777, + "grad_norm": 0.05841516703367233, + "kept_token_count": 653.490625, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9893001794815064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010699813067913056, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.6637237548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.79595947265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14390, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.1996454820036888, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.11867896467447281, + "grad_norm": 0.0871223658323288, + "kept_token_count": 654.85, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899482250213623, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010051767341792583, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1164978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.84185180664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14400, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.19936197251081467, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.12438562884926796, + "grad_norm": 0.054727908223867416, + "kept_token_count": 655.0390625, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902072966098785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009792698081582785, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.81723937988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5137481689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14410, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.20044663399457932, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.11335072666406631, + "grad_norm": 0.04784076288342476, + "kept_token_count": 654.1078125, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897342503070832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010265759099274874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0297424316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.23101806640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14420, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19938193261623383, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.13334558308124542, + "grad_norm": 0.0747869536280632, + "kept_token_count": 654.915625, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890909612178802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01090903701260686, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.6823211669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.66583251953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14430, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.19986813366413117, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.1236519306898117, + "grad_norm": 0.051757268607616425, + "kept_token_count": 654.6953125, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.990236121416092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009763876348733902, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.9017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.61806335449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14440, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2005625307559967, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.11803986057639122, + "grad_norm": 0.05574261397123337, + "kept_token_count": 654.2546875, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903772175312042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009622799698263407, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9505554199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4033447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14450, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2015366718173027, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.11316775679588317, + "grad_norm": 0.044743459671735764, + "kept_token_count": 653.03125, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904763460159302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009523659758269788, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5118103027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19359436035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14460, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.1994541436433792, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.12092535942792892, + "grad_norm": 0.07102248072624207, + "kept_token_count": 655.2453125, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.98952214717865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010477836709469557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.5472717285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2456970214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14470, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20072525143623351, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.11953299939632415, + "grad_norm": 0.050725650042295456, + "kept_token_count": 654.2640625, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901009261608124, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009899070113897323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.22984008789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3246276855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14480, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.19969191551208496, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.1279580108821392, + "grad_norm": 0.041968077421188354, + "kept_token_count": 654.80625, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9892156660556793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010784321743994952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.2716003417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.807861328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14490, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19962972551584243, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.11597944796085358, + "grad_norm": 0.05305822938680649, + "kept_token_count": 655.0890625, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989901727437973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010098261013627052, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.2229919433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.23854064941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14500, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20001830160617828, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.12509505525231362, + "grad_norm": 0.05286375433206558, + "kept_token_count": 654.6171875, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895146250724792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010485366452485323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.2095245361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.53941650390624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14510, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19850085079669952, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.12907603830099107, + "grad_norm": 0.07067709416151047, + "kept_token_count": 656.034375, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897764682769775, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010223555378615856, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.1953155517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.97939147949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14520, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.2004562646150589, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.1252252072095871, + "grad_norm": 0.04551582783460617, + "kept_token_count": 653.8765625, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898017585277558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010198254603892565, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.70020751953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.86998291015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14530, + "valid_token_count": 817.80625 + }, + { + "actual_token_mask_ratio": 0.20085735768079757, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.12156082838773727, + "grad_norm": 0.04918709769845009, + "kept_token_count": 653.9421875, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904101848602295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009589794930070639, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.7723114013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.70013122558595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14540, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.20005286931991578, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.13950954601168633, + "grad_norm": 0.06485628336668015, + "kept_token_count": 654.559375, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895056903362274, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010494316276162863, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.5414245605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32065734863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14550, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2008294031023979, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.12716253846883774, + "grad_norm": 0.05031532794237137, + "kept_token_count": 653.8203125, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899489462375641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010051051434129477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.04998474121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2019866943359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14560, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19979509860277175, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.11279372349381447, + "grad_norm": 0.04367027059197426, + "kept_token_count": 654.7734375, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899397492408752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010060243587940932, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7178955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.55101013183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14570, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.200007863342762, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.12097934037446975, + "grad_norm": 0.05458840727806091, + "kept_token_count": 654.5875, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895046055316925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010495371092110873, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.1944854736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.72092895507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14580, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.1998648852109909, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.12345394566655159, + "grad_norm": 0.07091150432825089, + "kept_token_count": 654.7109375, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895645499229431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010435460787266492, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.2820983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.00579833984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14590, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20001808404922486, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.11276410296559333, + "grad_norm": 0.051885928958654404, + "kept_token_count": 654.571875, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907012701034545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009298731479793787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.64068908691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1737945556641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14600, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20081140249967575, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.1273629881441593, + "grad_norm": 0.06652623414993286, + "kept_token_count": 653.7375, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.989300686120987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010699314065277576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0202178955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6582733154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14610, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.20041650980710984, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.11990730315446854, + "grad_norm": 0.04994071647524834, + "kept_token_count": 654.1359375, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896461844444275, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010353815928101539, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.7737152099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.55723266601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14620, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.20002213418483733, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.11482643410563469, + "grad_norm": 0.042906880378723145, + "kept_token_count": 654.64375, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901500940322876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009849909506738187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.88893127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.32313842773436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.19997769296169282, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.13207632005214692, + "grad_norm": 0.05406581982970238, + "kept_token_count": 654.5875, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904608845710754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00953910369426012, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0743682861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35651245117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14640, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19947919696569444, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.12003487274050713, + "grad_norm": 0.05899030342698097, + "kept_token_count": 655.115625, + "learning_rate": 1.681600449102196e-05, + "loss": 0.011, + "mgd_cos_sim": 0.9889623522758484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011037650890648365, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.30799255371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.89459838867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2004279002547264, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.12289713323116302, + "grad_norm": 0.06627285480499268, + "kept_token_count": 654.3125, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895416378974915, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010458372998982668, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.24449768066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1911285400391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14660, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19880799502134322, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.12868625521659852, + "grad_norm": 0.050754353404045105, + "kept_token_count": 655.696875, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0107, + "mgd_cos_sim": 0.9892907619476319, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010709252674132586, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.763720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1753204345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14670, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.1994321510195732, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.12620414644479752, + "grad_norm": 0.05184276029467583, + "kept_token_count": 655.3078125, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895219385623932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010478063020855188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.54917297363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.449267578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14680, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.20029677152633668, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.12096972018480301, + "grad_norm": 0.05155547708272934, + "kept_token_count": 654.325, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901614964008332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009838495310395956, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.72947082519534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6548217773437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14690, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19911833554506303, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.13267527297139167, + "grad_norm": 0.07011868059635162, + "kept_token_count": 655.459375, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899180114269257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01008198456838727, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.5452087402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.76775512695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14700, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.2003768041729927, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.12605441585183144, + "grad_norm": 0.05480285733938217, + "kept_token_count": 654.4875, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897004663944244, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010299532301723957, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.81305236816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2804321289062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14710, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.1995839834213257, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.1279970623552799, + "grad_norm": 0.05147239938378334, + "kept_token_count": 655.334375, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.989606523513794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010393480025231838, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.19244384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.93169860839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14720, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.20003324896097183, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.12774685472249986, + "grad_norm": 0.05307632312178612, + "kept_token_count": 654.678125, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900922358036042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009907763078808784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4210266113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.60588989257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14730, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.1999279662966728, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.11924428790807724, + "grad_norm": 0.06462258100509644, + "kept_token_count": 654.4625, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989934504032135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010065499600023032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.16998596191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93555603027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14740, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.19965384155511856, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.12123969420790673, + "grad_norm": 0.048695143312215805, + "kept_token_count": 654.7265625, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.990066385269165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0099336220882833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0275146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1900665283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14750, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.1998581424355507, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.12201848775148391, + "grad_norm": 0.04541943222284317, + "kept_token_count": 655.103125, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899169325828552, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010083066113293172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.63825073242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.64142150878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14760, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.2005346029996872, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.12064968198537826, + "grad_norm": 0.05363339185714722, + "kept_token_count": 654.5265625, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903768539428711, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009623149037361145, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.6625915527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2919555664063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14770, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.19870926439762115, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.12864433377981185, + "grad_norm": 0.055317673832178116, + "kept_token_count": 655.621875, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9895198166370391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010480182059109212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.89527282714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8869934082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14780, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19919155389070511, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.11642584726214408, + "grad_norm": 0.04683879762887955, + "kept_token_count": 655.2734375, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896591663360595, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0103408289141953, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.72120361328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5533111572266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14790, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19988024830818177, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.1240560233592987, + "grad_norm": 0.045936863869428635, + "kept_token_count": 654.6, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903300523757934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009669945761561393, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0121154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9339904785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14800, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20053936690092086, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.12892819568514824, + "grad_norm": 0.058535411953926086, + "kept_token_count": 654.4375, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899115920066833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010088394209742547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.3311370849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7076873779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14810, + "valid_token_count": 818.5984375 + }, + { + "actual_token_mask_ratio": 0.20016016215085983, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.12028302475810052, + "grad_norm": 0.040528733283281326, + "kept_token_count": 654.6625, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904779732227326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00952203618362546, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.7148742675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6428466796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14820, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.19987210035324096, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.12246430143713952, + "grad_norm": 0.06918568164110184, + "kept_token_count": 654.8453125, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.011, + "mgd_cos_sim": 0.988961535692215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011038451828062535, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.3060638427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.69525451660155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14830, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20025825649499893, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.12353517860174179, + "grad_norm": 0.06392791122198105, + "kept_token_count": 654.228125, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899433374404907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010056668519973755, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.5734161376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.68934631347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14840, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.19994679391384124, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.10773918479681015, + "grad_norm": 0.04661142826080322, + "kept_token_count": 654.6484375, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896517693996429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010348242148756981, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.0774200439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9545867919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14850, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20082727521657945, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.12223270237445831, + "grad_norm": 0.050387050956487656, + "kept_token_count": 654.046875, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904956638813018, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009504326339811086, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.0365875244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9932434082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14860, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.1996353879570961, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.12359722480177879, + "grad_norm": 0.05407441779971123, + "kept_token_count": 654.9546875, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989915382862091, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010084619373083114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0137969970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.18565979003904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14870, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.19935785681009294, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.12435575798153878, + "grad_norm": 0.05257158726453781, + "kept_token_count": 655.0953125, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900538086891174, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00994618246331811, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.82066955566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.260400390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14880, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20043513625860215, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.11718998327851296, + "grad_norm": 0.04087081551551819, + "kept_token_count": 654.44375, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.990355783700943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009644215740263463, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.66053466796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.986669921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14890, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.19934366196393966, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.12946312576532365, + "grad_norm": 0.06488750129938126, + "kept_token_count": 655.3421875, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901629328727722, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009837051760405302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.58790893554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53929138183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14900, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.200004243850708, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.12125936821103096, + "grad_norm": 0.05262655392289162, + "kept_token_count": 654.4890625, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.01, + "mgd_cos_sim": 0.98996621966362, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010033780895173549, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.9913269042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7072814941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.1996201053261757, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.12035970613360406, + "grad_norm": 0.05736410617828369, + "kept_token_count": 654.90625, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.01, + "mgd_cos_sim": 0.99002605676651, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009973942209035159, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3561218261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2620910644531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14920, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19981317073106766, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.13518389984965323, + "grad_norm": 0.04826432839035988, + "kept_token_count": 654.778125, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894497931003571, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01055020485073328, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3220581054687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70363159179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14930, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20026147961616517, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.11883114427328109, + "grad_norm": 0.07091197371482849, + "kept_token_count": 654.4953125, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900268018245697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009973200783133507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.22040100097655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.40911560058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14940, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2002980515360832, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.11679168120026588, + "grad_norm": 0.05036625638604164, + "kept_token_count": 654.3953125, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896048963069916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0103951096534729, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.63768310546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.96135864257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14950, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20027793049812317, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.11751971021294594, + "grad_norm": 0.042724061757326126, + "kept_token_count": 654.60625, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900621592998504, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009937840048223734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.1551239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4512420654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14960, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19986378401517868, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.12519318386912345, + "grad_norm": 0.05625443160533905, + "kept_token_count": 655.121875, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9894264042377472, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01057359278202057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.267529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1638214111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14970, + "valid_token_count": 818.7609375 + }, + { + "actual_token_mask_ratio": 0.20095379799604415, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.12227950170636177, + "grad_norm": 0.03790317475795746, + "kept_token_count": 653.6375, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904896080493927, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009510386548936368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.85109252929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3745635986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14980, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20026653707027436, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.11816525384783745, + "grad_norm": 0.05240730941295624, + "kept_token_count": 654.353125, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904731154441834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009526881761848927, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.18707885742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9788330078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14990, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20004747211933135, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.11974490582942962, + "grad_norm": 0.041794031858444214, + "kept_token_count": 654.6703125, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902600169181823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009739994164556265, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.8354522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93211669921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15000, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.19894497096538544, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.12219329476356507, + "grad_norm": 0.05491067096590996, + "kept_token_count": 655.6578125, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900852859020233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009914699476212263, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.58794555664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2688690185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15010, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.20093445032835006, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.122175682336092, + "grad_norm": 0.05402050167322159, + "kept_token_count": 653.9421875, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.990128630399704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009871364384889603, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.10682067871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.88689575195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15020, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.19994271844625472, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.11374636888504028, + "grad_norm": 0.05049188807606697, + "kept_token_count": 654.7015625, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901132881641388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00988671388477087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1345275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57969970703124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15030, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2000248059630394, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.11881310418248177, + "grad_norm": 0.08591563999652863, + "kept_token_count": 654.6703125, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902650833129882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009734922368079425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.8104614257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1030242919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19952163249254226, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.12156055495142937, + "grad_norm": 0.07472051680088043, + "kept_token_count": 655.1421875, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904940605163575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00950594674795866, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.78663940429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2261260986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15050, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.19964469373226165, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.12656320109963418, + "grad_norm": 0.06024911254644394, + "kept_token_count": 655.0390625, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9913160860538482, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008683910127729177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.51520080566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2587005615234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15060, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.20030054003000258, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.13189612179994584, + "grad_norm": 0.05476640909910202, + "kept_token_count": 654.196875, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902998626232147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009700131043791771, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5257171630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.25880737304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15070, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.2013409063220024, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.13591977208852768, + "grad_norm": 0.05695205181837082, + "kept_token_count": 653.6109375, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901657640933991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009834246337413787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.01650085449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.31871643066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15080, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.1999107614159584, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.11890568062663079, + "grad_norm": 0.0557372123003006, + "kept_token_count": 654.578125, + "learning_rate": 1.609197673279132e-05, + "loss": 0.01, + "mgd_cos_sim": 0.990039074420929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009960927627980709, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2789794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9778717041016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15090, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.199206081032753, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.12053466960787773, + "grad_norm": 0.05858964845538139, + "kept_token_count": 655.153125, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903824090957641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009617589227855206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.82526245117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4191955566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15100, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19992938935756682, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.1305051065981388, + "grad_norm": 0.046903178095817566, + "kept_token_count": 654.475, + "learning_rate": 1.605899245337355e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899682521820068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010031742881983519, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.9544372558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5206787109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15110, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.1999453127384186, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.11913537383079528, + "grad_norm": 0.05373081937432289, + "kept_token_count": 654.846875, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906358480453491, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009364175144582987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8882568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7370269775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15120, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20030632317066194, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.11560115739703178, + "grad_norm": 0.04252810776233673, + "kept_token_count": 654.5921875, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898173987865448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01018260046839714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5712493896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.71893920898435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15130, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.20002900213003158, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.12171501442790031, + "grad_norm": 0.0671781525015831, + "kept_token_count": 654.7640625, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.9894702553749084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010529738944023847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.3508270263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53843994140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15140, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.20000815242528916, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.11164687052369118, + "grad_norm": 0.047946661710739136, + "kept_token_count": 654.4015625, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901766240596771, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009823388513177633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0684387207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7540771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15150, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.20053666830062866, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.11776769533753395, + "grad_norm": 0.0790562629699707, + "kept_token_count": 653.93125, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893917620182038, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01060824878513813, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.13365783691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4839172363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15160, + "valid_token_count": 817.965625 + }, + { + "actual_token_mask_ratio": 0.20131638497114182, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.13145029321312904, + "grad_norm": 0.05862581729888916, + "kept_token_count": 653.39375, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.98957759141922, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010422401875257493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.3885986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9433074951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15170, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19977160692214965, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.12484624683856964, + "grad_norm": 0.051060765981674194, + "kept_token_count": 654.971875, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900346517562866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009965340606868267, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.1191162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3688262939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15180, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.1996582880616188, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.1353796660900116, + "grad_norm": 0.05248463526368141, + "kept_token_count": 654.8640625, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895723640918732, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010427631624042987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.25774536132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41676940917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.20005402266979216, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.12313364669680596, + "grad_norm": 0.04680207744240761, + "kept_token_count": 654.4015625, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901691854000092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009830821119248868, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.8248565673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10747680664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15200, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.20040552765130998, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.12352586388587952, + "grad_norm": 0.046341992914676666, + "kept_token_count": 654.434375, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903678834438324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009632123913615943, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.6650115966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.04572143554685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15210, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.19997478425502777, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.12655943632125854, + "grad_norm": 0.05504930019378662, + "kept_token_count": 654.415625, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897863507270813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010213651787489652, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.4536102294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6803009033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15220, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.19995268285274506, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.12655561789870262, + "grad_norm": 0.041986286640167236, + "kept_token_count": 654.6625, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905118644237518, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009488146286457777, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.48675842285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.75452880859376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15230, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.1992581680417061, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.12591999247670174, + "grad_norm": 0.06610118597745895, + "kept_token_count": 655.3171875, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897548913955688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010245109349489212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.73411254882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6633056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20025427639484406, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.11358650550246238, + "grad_norm": 0.04721352085471153, + "kept_token_count": 654.209375, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900623857975006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009937620535492898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.29473571777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.27460327148435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15250, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19945344775915147, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.10953832566738128, + "grad_norm": 0.06542132794857025, + "kept_token_count": 654.965625, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.990096265077591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009903745632618666, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4170684814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.3840789794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15260, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.2004494696855545, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.13234730958938598, + "grad_norm": 0.04307614266872406, + "kept_token_count": 654.1921875, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897463083267212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010253683011978865, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.2822967529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7057250976562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15270, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19998629093170167, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.12357092201709748, + "grad_norm": 0.04843078926205635, + "kept_token_count": 654.71875, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902268290519715, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009773177560418844, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.40984497070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8896423339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15280, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19998517036437988, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.12159310206770897, + "grad_norm": 0.04937784746289253, + "kept_token_count": 654.584375, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904764771461487, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0095235382206738, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2515045166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.91908569335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15290, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.19969026148319244, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.1287938393652439, + "grad_norm": 0.06445952504873276, + "kept_token_count": 654.88125, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898057460784913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010194245632737875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.5255432128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.25267639160154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15300, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19954869151115417, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.12419586330652237, + "grad_norm": 0.06977572292089462, + "kept_token_count": 655.009375, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0112, + "mgd_cos_sim": 0.988815575838089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.011184433195739984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3441497802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1171844482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15310, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20077465921640397, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.12106463611125946, + "grad_norm": 0.048029910773038864, + "kept_token_count": 654.0515625, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.9891854643821716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010814528539776802, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.04564208984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7843841552734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15320, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20011416971683502, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.1294682316482067, + "grad_norm": 0.05854429677128792, + "kept_token_count": 654.2859375, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900777339935303, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009922277089208365, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.15667114257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8911102294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15330, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.20004921406507492, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.12516747787594795, + "grad_norm": 0.05525994300842285, + "kept_token_count": 654.5640625, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907882571220398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009211743529886008, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4407684326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4711456298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15340, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.20073706060647964, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.13120789527893068, + "grad_norm": 0.045500583946704865, + "kept_token_count": 653.928125, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903847217559815, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009615294728428125, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9713592529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.91387939453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15350, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.1997988685965538, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.12926326170563698, + "grad_norm": 0.0529499389231205, + "kept_token_count": 655.0140625, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907507121562957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009249294176697731, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7834503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4897216796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15360, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.19938876777887343, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.13277650773525237, + "grad_norm": 0.05592994764447212, + "kept_token_count": 655.2046875, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898929178714753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010107067134231328, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1000274658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.98296508789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15370, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.19929028153419495, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.11407468542456627, + "grad_norm": 0.06373787671327591, + "kept_token_count": 655.0953125, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901142001152039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009885791130363942, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.0618560791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4866455078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15380, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.1994596615433693, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.12147672474384308, + "grad_norm": 0.06118306517601013, + "kept_token_count": 655.04375, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899032354354859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010096760839223862, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.84649353027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4035125732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15390, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2002124473452568, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.12493554204702377, + "grad_norm": 0.0490647628903389, + "kept_token_count": 654.690625, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904739618301391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009526039101183414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.99521789550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.91563720703124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15400, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.2007622852921486, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.11950928941369057, + "grad_norm": 0.057885102927684784, + "kept_token_count": 653.9546875, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900699079036712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009930099546909332, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7059631347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0669372558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15410, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.1995356336236, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.13281560316681862, + "grad_norm": 0.04106087237596512, + "kept_token_count": 655.1046875, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904653370380402, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00953466035425663, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.7554046630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8044891357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15420, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.19990203827619552, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.11688343435525894, + "grad_norm": 0.0592489019036293, + "kept_token_count": 654.778125, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903481066226959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009651883598417043, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1008636474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8462341308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15430, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.19960129410028457, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.12605829015374184, + "grad_norm": 0.04611865058541298, + "kept_token_count": 655.0859375, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901875257492065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009812471829354763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.9478240966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5715087890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15440, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.1993141531944275, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.1152000293135643, + "grad_norm": 0.05101734772324562, + "kept_token_count": 655.290625, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895901024341583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010409898310899734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.97258911132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53246154785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15450, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.1998363584280014, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.12659662440419198, + "grad_norm": 0.05894199013710022, + "kept_token_count": 654.625, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901975512504577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009802462067455053, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3825714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.60487976074216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15460, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.1999232903122902, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.11716196909546853, + "grad_norm": 0.04415445029735565, + "kept_token_count": 654.74375, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903191685676574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009680826682597399, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.28058166503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8351257324219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19981496930122375, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.13098858445882797, + "grad_norm": 0.04437483474612236, + "kept_token_count": 654.3171875, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903710365295411, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009628957137465477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.42274780273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7007171630859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15480, + "valid_token_count": 817.709375 + }, + { + "actual_token_mask_ratio": 0.2011359453201294, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.12630755677819253, + "grad_norm": 0.06578550487756729, + "kept_token_count": 653.5765625, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902296423912048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009770347829908132, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.0502136230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5666564941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15490, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19954636991024016, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.12287476882338524, + "grad_norm": 0.0607156939804554, + "kept_token_count": 654.865625, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900819301605225, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009918079804629087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4025085449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.56410217285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15500, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.1996999204158783, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.1244470164179802, + "grad_norm": 0.09371557831764221, + "kept_token_count": 655.0515625, + "learning_rate": 1.539843935650351e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900078296661377, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009992173407226801, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2753204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.31561889648435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15510, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20123166292905809, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.13419144749641418, + "grad_norm": 0.1015821322798729, + "kept_token_count": 653.4640625, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902620315551758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009737960994243622, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.023095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3538421630859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15520, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2002235770225525, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.12793960571289062, + "grad_norm": 0.07876794040203094, + "kept_token_count": 654.325, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910623013973237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008937697950750589, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.75486145019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3210845947266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15530, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.1999960094690323, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.12327903062105179, + "grad_norm": 0.08654125034809113, + "kept_token_count": 654.26875, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0105, + "mgd_cos_sim": 0.989525580406189, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010474426299333572, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7569915771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.43963012695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15540, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.20039335191249846, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.12721766382455826, + "grad_norm": 0.04350833222270012, + "kept_token_count": 654.5703125, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905027568340301, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009497243724763394, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.62377014160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.526416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15550, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.20082437992095947, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.1289341039955616, + "grad_norm": 0.06324565410614014, + "kept_token_count": 653.7875, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904159009456635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009584088064730167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8662353515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.79730224609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15560, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.20007964521646499, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.12943490743637084, + "grad_norm": 0.05414275452494621, + "kept_token_count": 654.575, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898668229579926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010133186168968678, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.7596160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41860046386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15570, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20018069595098495, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.12773538157343864, + "grad_norm": 0.04396110773086548, + "kept_token_count": 654.9015625, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902516663074493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00974832996726036, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5565246582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15580, + "valid_token_count": 818.815625 + }, + { + "actual_token_mask_ratio": 0.19969518333673478, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.11315756216645241, + "grad_norm": 0.052544329315423965, + "kept_token_count": 654.9265625, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901403665542603, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009859633166342974, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0506561279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15590, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.20049954056739808, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.12472798824310302, + "grad_norm": 0.061520084738731384, + "kept_token_count": 654.490625, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9896714746952057, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010328530799597502, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.4986877441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.42506713867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15600, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20102487206459047, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.12930630445480346, + "grad_norm": 0.06231934204697609, + "kept_token_count": 654.090625, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902076840400695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009792305901646614, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8529815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0288055419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15610, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.19992576092481612, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.11800113767385483, + "grad_norm": 0.0640033408999443, + "kept_token_count": 654.7015625, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900712370872498, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009928758069872857, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.878369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.017431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15620, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2006731316447258, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.11951364949345589, + "grad_norm": 0.05577569827437401, + "kept_token_count": 654.1, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903064787387847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009693525824695825, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.85126037597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.34104919433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15630, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.20037320107221604, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.12754328772425652, + "grad_norm": 0.06294789910316467, + "kept_token_count": 654.5015625, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.989906269311905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010093737952411174, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.80230712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.62705688476564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15640, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.20094497203826905, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.11934189721941948, + "grad_norm": 0.08265918493270874, + "kept_token_count": 653.8375, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900008916854859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00999910682439804, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7166381835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.78150329589846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15650, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20018836259841918, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.12665611505508423, + "grad_norm": 0.0721217542886734, + "kept_token_count": 654.6703125, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898650765419006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010134924296289682, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.03505554199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.92316284179685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15660, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.19971641302108764, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.12045988738536835, + "grad_norm": 0.06652826815843582, + "kept_token_count": 655.05625, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899275779724122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010072405822575093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.39222412109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.10215759277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15670, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.19972513616085052, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.12437927573919297, + "grad_norm": 0.05039479210972786, + "kept_token_count": 654.8, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901639223098755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009836067538708448, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.18328552246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.18271789550784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19948429018259048, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.12267508134245872, + "grad_norm": 0.05726224184036255, + "kept_token_count": 655.00625, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.990201449394226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009798556566238403, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7872344970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7338012695312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15690, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19992629140615464, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.12998261600732802, + "grad_norm": 0.06072527915239334, + "kept_token_count": 654.5421875, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.98920179605484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010798199381679297, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3798370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4855194091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15700, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20032216608524323, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.12217336297035217, + "grad_norm": 0.0504373237490654, + "kept_token_count": 654.115625, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904222428798676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009577754512429237, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9287994384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.33522338867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15710, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.20100803375244142, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.12288499474525452, + "grad_norm": 0.06893587112426758, + "kept_token_count": 653.85, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903487443923951, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009651260543614626, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9866119384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3009063720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2004665344953537, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.12487594857811928, + "grad_norm": 0.05866728350520134, + "kept_token_count": 654.2875, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899821400642395, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010017870739102364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.49058227539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2837219238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15730, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.20025035738945007, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.118509890884161, + "grad_norm": 0.06753742694854736, + "kept_token_count": 654.278125, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898472547531127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010152740217745305, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1009094238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8135620117188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15740, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19974946975708008, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.12577340826392175, + "grad_norm": 0.04455801472067833, + "kept_token_count": 654.55625, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906846165657044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0093153677880764, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4583343505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.96328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15750, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.20045101195573806, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.12126951515674592, + "grad_norm": 0.06842681765556335, + "kept_token_count": 654.1984375, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0109, + "mgd_cos_sim": 0.9890837132930755, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010916284192353487, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9206848144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.75050659179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15760, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19953546822071075, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.1285709485411644, + "grad_norm": 0.055730607360601425, + "kept_token_count": 654.9296875, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.990404611825943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009595379140228033, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4358856201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.65472106933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.1998216226696968, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.12795535549521447, + "grad_norm": 0.04811016470193863, + "kept_token_count": 654.51875, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.990353274345398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009646718669682742, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8124542236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5508514404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15780, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.19982582777738572, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.12790868133306504, + "grad_norm": 0.053393296897411346, + "kept_token_count": 654.740625, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906060874462128, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00939390528947115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.18504028320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.04505920410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20021494776010512, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.11677579209208488, + "grad_norm": 0.05905720591545105, + "kept_token_count": 654.675, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899302124977112, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010069793369621039, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.9206848144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2700988769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15800, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.20011138319969177, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.12386688068509102, + "grad_norm": 0.0555657260119915, + "kept_token_count": 654.371875, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.990353137254715, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009646868705749512, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.577490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.85089721679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15810, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20042258948087693, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.12131982818245887, + "grad_norm": 0.05781970173120499, + "kept_token_count": 654.1640625, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901708006858826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009829198569059372, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3786865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.54534606933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15820, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19945882111787797, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.11765859872102738, + "grad_norm": 0.0426613874733448, + "kept_token_count": 655.06875, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908910036087036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009108994249254465, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0978088378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.24322509765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15830, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2001022920012474, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.12494666129350662, + "grad_norm": 0.07450103759765625, + "kept_token_count": 654.546875, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0106, + "mgd_cos_sim": 0.9893839836120606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010616021137684584, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.41119384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2908416748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15840, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20082345753908157, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.12402926683425904, + "grad_norm": 0.05287090316414833, + "kept_token_count": 653.890625, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906647264957428, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00933528235182166, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.64466247558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83404846191405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15850, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.20049877613782882, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.12264011651277543, + "grad_norm": 0.0501568578183651, + "kept_token_count": 654.25625, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902153074741363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009784699883311987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.6856964111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.95811767578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15860, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2000029966235161, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.12173385694622993, + "grad_norm": 0.06454917788505554, + "kept_token_count": 654.5375, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905907571315765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009409235138446093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9668731689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2169128417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15870, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20018427819013596, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.12269600331783295, + "grad_norm": 0.062331318855285645, + "kept_token_count": 654.478125, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.990474647283554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00952536128461361, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.5594421386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1441619873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15880, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.1998894676566124, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.12559668570756913, + "grad_norm": 0.04466928169131279, + "kept_token_count": 654.409375, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904012501239776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00959874400869012, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.15550231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3585723876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15890, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.200584015250206, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.12255982682108879, + "grad_norm": 0.047449205070734024, + "kept_token_count": 654.0140625, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901880085468292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009811983536928892, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.13902893066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0969970703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15900, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.20081968903541564, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.12068376541137696, + "grad_norm": 0.06293976306915283, + "kept_token_count": 654.1984375, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905424594879151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009457544144243001, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4148864746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19313354492186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15910, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.19946991503238679, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.12777801156044005, + "grad_norm": 0.04456121474504471, + "kept_token_count": 655.2765625, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898817002773285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010118303261697293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.5697448730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.38067321777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15920, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.20050398856401444, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.12051356062293053, + "grad_norm": 0.06992311030626297, + "kept_token_count": 654.3921875, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0108, + "mgd_cos_sim": 0.989189499616623, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010810503084212542, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.43817749023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.73586120605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15930, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19964411109685898, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.11727606281638145, + "grad_norm": 0.06768149137496948, + "kept_token_count": 655.0890625, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909464359283447, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009053571894764901, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.42674560546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0687957763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15940, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20014770925045014, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.12399591058492661, + "grad_norm": 0.061969418078660965, + "kept_token_count": 654.628125, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908127665519715, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009187221806496382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.6752685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70567016601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15950, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.20025366246700288, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.12492372915148735, + "grad_norm": 0.048183441162109375, + "kept_token_count": 654.5765625, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.990155839920044, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009844159614294768, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8556213378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4032287597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15960, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.19942645579576493, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.12426568195223808, + "grad_norm": 0.052420955151319504, + "kept_token_count": 655.2984375, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.991224217414856, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008775782398879528, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7786560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3390380859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15970, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.2006011947989464, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.12882919758558273, + "grad_norm": 0.051656898111104965, + "kept_token_count": 654.121875, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901144504547119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009885549079626799, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.70823364257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9151916503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15980, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19973333030939103, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.1252674363553524, + "grad_norm": 0.05674486607313156, + "kept_token_count": 654.9515625, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9895874798297882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010412514209747314, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.69191284179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6355682373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15990, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20061888545751572, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.11886783167719842, + "grad_norm": 0.05830811709165573, + "kept_token_count": 654.3640625, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.990264070034027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009735941328108311, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.25010070800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.08542785644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16000, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.20063843578100204, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.11751258894801139, + "grad_norm": 0.0480157844722271, + "kept_token_count": 653.90625, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899507641792298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010049234330654144, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.3264434814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1953521728516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16010, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.1994728446006775, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.13140744864940643, + "grad_norm": 0.05892352759838104, + "kept_token_count": 654.8828125, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899950385093689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010004970710724592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9231231689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.23781127929686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16020, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.19925057888031006, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.11567162796854973, + "grad_norm": 0.04961920529603958, + "kept_token_count": 655.4171875, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904822289943696, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009517772868275643, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6363128662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3117950439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16030, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.19912792295217513, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.1291039727628231, + "grad_norm": 0.059577811509370804, + "kept_token_count": 655.3890625, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904456436634064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009554353635758161, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3970031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.81105651855466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16040, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19965083599090577, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.12793745025992392, + "grad_norm": 0.04485632851719856, + "kept_token_count": 654.7953125, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903089821338653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009691010322421789, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0111114501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6422180175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.1998114675283432, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.12371878400444984, + "grad_norm": 0.07709161192178726, + "kept_token_count": 654.515625, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906064808368683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00939351161941886, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.51585083007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.51344299316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16060, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.2001437395811081, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.13051470890641212, + "grad_norm": 0.06553389132022858, + "kept_token_count": 654.621875, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902536988258361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009746288880705833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0282867431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.75221557617186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.20043295472860337, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.11749653294682502, + "grad_norm": 0.056409768760204315, + "kept_token_count": 654.0890625, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905761897563934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009423802327364682, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.52696228027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.197607421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16080, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.199538753926754, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.12679583951830864, + "grad_norm": 0.05227954313158989, + "kept_token_count": 654.75, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906952083110809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0093048014678061, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0791809082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0467803955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16090, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.19997094422578812, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.1129389263689518, + "grad_norm": 0.042387690395116806, + "kept_token_count": 654.628125, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898862898349762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010113705601543187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.92158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.63807067871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16100, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19952063858509064, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.12327956557273864, + "grad_norm": 0.04777617007493973, + "kept_token_count": 654.840625, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.99039705991745, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00960294734686613, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.59861755371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1169677734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16110, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.19957851469516755, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.11960614398121834, + "grad_norm": 0.04751205816864967, + "kept_token_count": 654.9859375, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901334822177887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009866505395621061, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.32919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.91253662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16120, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2002976953983307, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.12399935647845269, + "grad_norm": 0.08309116959571838, + "kept_token_count": 654.290625, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899632811546326, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010036712698638438, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.42010803222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0583557128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16130, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20051850974559784, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.12773448079824448, + "grad_norm": 0.06453539431095123, + "kept_token_count": 654.128125, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902509987354279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009749004617333412, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.6915313720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01206665039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16140, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2000368356704712, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.12354362532496452, + "grad_norm": 0.05414234474301338, + "kept_token_count": 654.175, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901076018810272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009892396070063115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.75224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5701568603516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16150, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.1997709020972252, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.12493250966072082, + "grad_norm": 0.04797406122088432, + "kept_token_count": 654.5390625, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990932685136795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009067324921488762, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0198120117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7994812011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16160, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.19965997189283372, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.12343167886137962, + "grad_norm": 0.051565080881118774, + "kept_token_count": 655.0421875, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904847383499146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009515261091291905, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.92881774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.65924682617185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20019188672304153, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.1310948520898819, + "grad_norm": 0.07108699530363083, + "kept_token_count": 654.515625, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9898054242134094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010194563306868076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5479370117188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8917633056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16180, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19904048144817352, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.12719050720334052, + "grad_norm": 0.0512479692697525, + "kept_token_count": 655.40625, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990905374288559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009094622172415256, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.9475952148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8627655029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16190, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.2009797841310501, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.12954443022608758, + "grad_norm": 0.051414236426353455, + "kept_token_count": 653.6953125, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901004552841186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009899541269987822, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.6713623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.20408630371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16200, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19979167431592942, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.1318306230008602, + "grad_norm": 0.0527656264603138, + "kept_token_count": 654.7875, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0104, + "mgd_cos_sim": 0.9896139562129974, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010386042390018701, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.30901489257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49782409667966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16210, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1982800170779228, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.13040608540177345, + "grad_norm": 0.0548614040017128, + "kept_token_count": 655.9578125, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907276213169098, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009272382967174054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5368316650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3066833496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16220, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20057438760995866, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.1169903188943863, + "grad_norm": 0.047552596777677536, + "kept_token_count": 654.096875, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907597482204438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009240244887769223, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7076873779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.80787963867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.1999681919813156, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.11885858699679375, + "grad_norm": 0.04678913950920105, + "kept_token_count": 654.865625, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907522320747375, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00924777053296566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6344909667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3084289550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16240, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.19931062906980515, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.124915711581707, + "grad_norm": 0.0503535121679306, + "kept_token_count": 655.2984375, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908334970474243, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009166503790766, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.92929992675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.53163452148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16250, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19982047230005265, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.1252648837864399, + "grad_norm": 0.046445079147815704, + "kept_token_count": 654.6203125, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.990407544374466, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009592463821172714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.66832275390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.41655883789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16260, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.19974642843008042, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.12976874858140947, + "grad_norm": 0.05393172800540924, + "kept_token_count": 655.0015625, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906145393848419, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009385471511632205, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6924255371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93170166015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16270, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.2002526953816414, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.12162751331925392, + "grad_norm": 0.037778157740831375, + "kept_token_count": 654.53125, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901322841644287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009867713693529368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8092315673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.36387939453124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16280, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20044018179178238, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.13035957291722297, + "grad_norm": 0.04475503787398338, + "kept_token_count": 654.215625, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904120802879334, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009587909746915101, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1614685058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9853790283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19952018558979034, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.12216328755021096, + "grad_norm": 0.04696834832429886, + "kept_token_count": 655.2859375, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905882775783539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009411736018955707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.05361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.07020568847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16300, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.2007163092494011, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.11568100303411484, + "grad_norm": 0.04926285892724991, + "kept_token_count": 653.8453125, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907446265220642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00925536947324872, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.07074279785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6933258056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16310, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.19979146420955657, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.13242753893136977, + "grad_norm": 0.06379015743732452, + "kept_token_count": 654.9125, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904412508010865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009558745753020048, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4532928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0084228515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16320, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20015523582696915, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.11420216262340546, + "grad_norm": 0.1041981428861618, + "kept_token_count": 654.5546875, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909009814262391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009099029656499624, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.883203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.52548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16330, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19984900951385498, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.13150388300418853, + "grad_norm": 0.05084272101521492, + "kept_token_count": 654.625, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901305139064789, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009869495872408152, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.54600524902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87603149414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16340, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.1995275855064392, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.11866707280278206, + "grad_norm": 0.052124228328466415, + "kept_token_count": 654.9671875, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906325459480285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00936745386570692, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7943420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7758026123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16350, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19958883970975877, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.12944263815879822, + "grad_norm": 0.08148083835840225, + "kept_token_count": 654.86875, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902727842330933, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009727203194051981, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.69775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9811676025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16360, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19940026253461837, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.11891242936253547, + "grad_norm": 0.051250968128442764, + "kept_token_count": 655.1, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908959150314331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00910406829789281, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.62340393066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.15907287597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16370, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19976830929517747, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.12590288147330284, + "grad_norm": 0.04710714519023895, + "kept_token_count": 654.7578125, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905341565608978, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009465845674276352, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6458801269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1317962646484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16380, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20070720314979554, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.12573998272418976, + "grad_norm": 0.0703103095293045, + "kept_token_count": 654.0015625, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897676289081574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010232368670403957, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.4278839111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.63267211914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16390, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19999324679374694, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.12581930458545684, + "grad_norm": 0.042433325201272964, + "kept_token_count": 654.53125, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907887399196624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009211257006973029, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.9374145507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.43329467773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16400, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20051780641078948, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.12177886590361595, + "grad_norm": 0.06184260919690132, + "kept_token_count": 654.225, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990920615196228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009079376328736543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.44166870117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.42217407226565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16410, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2000126510858536, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.11584630534052849, + "grad_norm": 0.04454546421766281, + "kept_token_count": 654.6140625, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905070126056671, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009492980781942606, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4594207763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.48467712402345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.20053637474775315, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.11433420926332474, + "grad_norm": 0.042480770498514175, + "kept_token_count": 654.3609375, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907618641853333, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00923811485990882, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.24236450195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8212524414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16430, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.20072034746408463, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.11807349175214768, + "grad_norm": 0.04297131299972534, + "kept_token_count": 654.1390625, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0084, + "mgd_cos_sim": 0.9916330635547638, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008366949763149023, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8465179443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.99000549316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16440, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20057158023118973, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.12434452399611473, + "grad_norm": 0.04993487522006035, + "kept_token_count": 654.090625, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899118542671204, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01008813763037324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9881896972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.85797119140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16450, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20060532689094543, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.11506672501564026, + "grad_norm": 0.04415015131235123, + "kept_token_count": 654.3484375, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908224999904632, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009177497774362563, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.26140747070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14124755859376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16460, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.19952608197927474, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.12922395169734954, + "grad_norm": 0.05387158319354057, + "kept_token_count": 655.2125, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906917154788971, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009308298118412495, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0872741699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.46751708984374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16470, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.19960546046495437, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.12783983573317528, + "grad_norm": 0.0528152696788311, + "kept_token_count": 654.9109375, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903601884841919, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009639817662537098, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8635498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8131103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16480, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19892746806144715, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.12255525141954422, + "grad_norm": 0.04754289612174034, + "kept_token_count": 655.6015625, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910183608531952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008981642127037049, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7346160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.923193359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16490, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.19989901036024094, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.11877120137214661, + "grad_norm": 0.05233217403292656, + "kept_token_count": 654.83125, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909981608390808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009001861978322267, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.720556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.17755432128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16500, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19972530603408814, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.12867761254310608, + "grad_norm": 0.05440565571188927, + "kept_token_count": 654.9875, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901496827602386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009850307181477547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.21857299804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55226745605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16510, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.20020408928394318, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.11728211492300034, + "grad_norm": 0.0446607805788517, + "kept_token_count": 654.359375, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912617802619934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008738232031464576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.65415649414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9144256591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16520, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20010747611522675, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.12496008276939392, + "grad_norm": 0.0787602886557579, + "kept_token_count": 654.584375, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906903862953186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009309598337858915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8979034423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0022888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16530, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2006632447242737, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.12026598230004311, + "grad_norm": 0.045995764434337616, + "kept_token_count": 654.096875, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900481402873993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00995185598731041, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.16448669433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.17861022949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20042897164821624, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.12398359850049019, + "grad_norm": 0.05361194536089897, + "kept_token_count": 654.4265625, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905448794364929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009455129597336054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3749694824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.30174255371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16550, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.20015774220228194, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.11795339062809944, + "grad_norm": 0.060727111995220184, + "kept_token_count": 654.528125, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906687259674072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009331261459738015, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.39214477539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.78847045898436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19988655000925065, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.12141852900385856, + "grad_norm": 0.04627373069524765, + "kept_token_count": 654.8625, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9901154935359955, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0098844975233078, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.07843627929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31911010742186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16570, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2001655027270317, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.12342562898993492, + "grad_norm": 0.06388192623853683, + "kept_token_count": 655.0859375, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905990183353424, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009400988183915614, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.18865966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.13084411621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16580, + "valid_token_count": 819.025 + }, + { + "actual_token_mask_ratio": 0.19927681535482406, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.12083054706454277, + "grad_norm": 0.04873579367995262, + "kept_token_count": 655.0140625, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907945454120636, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009205452725291252, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.7702941894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.242919921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16590, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19977863430976867, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.12719597518444062, + "grad_norm": 0.04373877868056297, + "kept_token_count": 655.071875, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.989775538444519, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01022446807473898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0935485839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.24818115234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16600, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.19959785789251328, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.12268733233213425, + "grad_norm": 0.09141476452350616, + "kept_token_count": 654.990625, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903849184513092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009615085367113353, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2326721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.10082092285154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16610, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.20097684860229492, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.12763296887278558, + "grad_norm": 0.05380455404520035, + "kept_token_count": 653.6921875, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903363943099975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009663594327867031, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0590484619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.21693725585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16620, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20016883462667465, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.12938810214400293, + "grad_norm": 0.04996078833937645, + "kept_token_count": 654.61875, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908357441425324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009164248406887055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.08643493652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.00262756347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16630, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.20052294433116913, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.1354292444884777, + "grad_norm": 0.044689979404211044, + "kept_token_count": 654.39375, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903052747249603, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009694725181907416, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4461395263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9588989257812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16640, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19866010695695877, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.1295697920024395, + "grad_norm": 0.04790421202778816, + "kept_token_count": 655.6484375, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904347836971283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009565205313265324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3468505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1517395019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16650, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19902995675802232, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.1262584574520588, + "grad_norm": 0.06186765432357788, + "kept_token_count": 655.51875, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908391833305359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009160810150206089, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.7951202392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8268280029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16660, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19971567541360855, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.13577753230929374, + "grad_norm": 0.05461595207452774, + "kept_token_count": 654.896875, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.990300852060318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009699138067662715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.39751586914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.14874267578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20031561851501464, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.1197941817343235, + "grad_norm": 0.05050574988126755, + "kept_token_count": 654.58125, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909879088401794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009012095723301172, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2535034179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.05555419921876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16680, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.1997236043214798, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.13053466975688935, + "grad_norm": 0.055254314094781876, + "kept_token_count": 654.95625, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904982030391694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009501788206398487, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.27256774902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9638702392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16690, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.1996629536151886, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.12635462060570718, + "grad_norm": 0.05253184214234352, + "kept_token_count": 655.071875, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907547116279602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009245290700346232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5337646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4836090087891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16700, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.19965185821056367, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.1198395885527134, + "grad_norm": 0.04789941757917404, + "kept_token_count": 655.0046875, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904069602489471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00959304366260767, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2045928955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8629943847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16710, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2001340791583061, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.11893011257052422, + "grad_norm": 0.048584651201963425, + "kept_token_count": 654.5359375, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905588924884796, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009441104717552662, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.53839111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25478210449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16720, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19966820478439332, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.12208768054842949, + "grad_norm": 0.07510683685541153, + "kept_token_count": 654.8734375, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899161458015442, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010083849262446164, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4185852050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.11387023925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16730, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20029275715351105, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.11909581869840621, + "grad_norm": 0.049727071076631546, + "kept_token_count": 654.6453125, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911974370479584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008802549913525581, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.62970275878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5610015869141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16740, + "valid_token_count": 818.609375 + }, + { + "actual_token_mask_ratio": 0.2005465269088745, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.1156439132988453, + "grad_norm": 0.0506049208343029, + "kept_token_count": 654.25625, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912132084369659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008786790631711483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.11638793945315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.37254333496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16750, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.19992930442094803, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.1315982945263386, + "grad_norm": 0.06121514365077019, + "kept_token_count": 655.0078125, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907596170902252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00924037378281355, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2770172119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5069946289062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16760, + "valid_token_count": 818.6890625 + }, + { + "actual_token_mask_ratio": 0.20127734243869783, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.12196264266967774, + "grad_norm": 0.05838755518198013, + "kept_token_count": 653.95625, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909392774105072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00906070927157998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7854522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.71168823242186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16770, + "valid_token_count": 818.7546875 + }, + { + "actual_token_mask_ratio": 0.20053122341632842, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.12220170721411705, + "grad_norm": 0.04385644569993019, + "kept_token_count": 654.3390625, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911672234535217, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008832772076129914, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7645538330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.97906494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16780, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20071433633565902, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.1166925460100174, + "grad_norm": 0.045801810920238495, + "kept_token_count": 653.94375, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9913455545902252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008654444478452206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.60329284667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7602203369141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16790, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1995681956410408, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.12392225712537766, + "grad_norm": 0.06649360805749893, + "kept_token_count": 655.0859375, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909597218036652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009040280245244503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.09373779296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.64508361816405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16800, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.19990851134061813, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.12274853885173798, + "grad_norm": 0.04240823909640312, + "kept_token_count": 654.675, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.991357147693634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008642847836017608, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2925994873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.01800537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19882135540246965, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.12487680241465568, + "grad_norm": 0.06616351753473282, + "kept_token_count": 655.5125, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909029424190521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009097048453986645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6465606689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.57208862304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16820, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2006501391530037, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.12915924862027167, + "grad_norm": 0.07055990397930145, + "kept_token_count": 654.0359375, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903937160968781, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009606290142983198, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.38208923339846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.16045837402345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16830, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.1996776804327965, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.1267335183918476, + "grad_norm": 0.0517115481197834, + "kept_token_count": 655.23125, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.009, + "mgd_cos_sim": 0.990964949131012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009035043232142926, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.41883239746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.01487426757814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16840, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.20092544853687286, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.12180501371622085, + "grad_norm": 0.053738370537757874, + "kept_token_count": 653.9609375, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905202031135559, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009479795303195714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.7888122558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2231079101563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20012924820184708, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.11470258831977845, + "grad_norm": 0.05794961005449295, + "kept_token_count": 654.46875, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9899838268756866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010016171541064978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.2845428466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.15748901367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16860, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19927683025598525, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.12060441896319389, + "grad_norm": 0.06385000795125961, + "kept_token_count": 655.2640625, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905486285686493, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00945135960355401, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.04425354003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3772399902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16870, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20000147670507432, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.12131966650485992, + "grad_norm": 0.06835614144802094, + "kept_token_count": 654.809375, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901791512966156, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009820857644081115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.47401123046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3901672363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16880, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.19857370853424072, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.11312040016055107, + "grad_norm": 0.05830727517604828, + "kept_token_count": 655.959375, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9913289546966553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008671031892299652, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.9695587158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0781951904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16890, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.19994594603776933, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.12204622402787209, + "grad_norm": 0.04341685399413109, + "kept_token_count": 654.4578125, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904850900173188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009514909982681275, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.1810272216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3527496337891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16900, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.19973899126052858, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.11666815057396888, + "grad_norm": 0.05961821973323822, + "kept_token_count": 655.2109375, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9914338231086731, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008566189277917146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4089660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3617828369141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16910, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.19970449507236482, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.1250500924885273, + "grad_norm": 0.052852947264909744, + "kept_token_count": 654.9015625, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910859227180481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008914073836058379, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6639373779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.011376953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16920, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.19981923401355745, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.12351437732577324, + "grad_norm": 0.06120225042104721, + "kept_token_count": 654.8625, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898769080638885, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010123111959546804, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6118865966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.82428588867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16930, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19962648153305054, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.11933762133121491, + "grad_norm": 0.05657719448208809, + "kept_token_count": 654.984375, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.990285986661911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0097140041179955, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.86224060058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3541717529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16940, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19976666271686555, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.12760545313358307, + "grad_norm": 0.05614367127418518, + "kept_token_count": 654.8484375, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903197169303894, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009680272545665503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4225646972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.67027893066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16950, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.1999959483742714, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.10878733098506928, + "grad_norm": 0.05354754254221916, + "kept_token_count": 654.6171875, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905708849430084, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009429108165204526, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.484765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.3150909423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16960, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19916862547397612, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.12510138228535653, + "grad_norm": 0.04371664673089981, + "kept_token_count": 655.4703125, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900038719177247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009996126778423787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0226531982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4356292724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16970, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.20110484957695007, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.13240943923592569, + "grad_norm": 0.04447735473513603, + "kept_token_count": 653.5671875, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.990317314863205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009682673402130604, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4997528076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0978240966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16980, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.199646058678627, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.12780174538493155, + "grad_norm": 0.04484325647354126, + "kept_token_count": 654.9640625, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906921744346618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009307813830673695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.00847778320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8572937011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16990, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19977097511291503, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.12839000076055526, + "grad_norm": 0.049744863063097, + "kept_token_count": 654.69375, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909294068813324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009070579707622529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3253875732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.00658264160154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17000, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19968172311782836, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.12173907607793807, + "grad_norm": 0.04555639624595642, + "kept_token_count": 654.7515625, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.990332055091858, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009667947329580784, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.64085083007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.42601623535154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17010, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19981074184179307, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.12112929821014404, + "grad_norm": 0.04734775796532631, + "kept_token_count": 654.8265625, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897705614566803, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010229435749351978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0056396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.46852722167966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17020, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20038582235574723, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.11783370971679688, + "grad_norm": 0.04174541309475899, + "kept_token_count": 654.203125, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908509910106659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009149012994021178, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8350341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5066619873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17030, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19901825487613678, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.12355516701936722, + "grad_norm": 0.10829033702611923, + "kept_token_count": 655.3234375, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9899061977863312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010093807987868787, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.752685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3952972412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17040, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.1993807315826416, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.1236785151064396, + "grad_norm": 0.06332293897867203, + "kept_token_count": 655.0625, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903411090373992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009658891893923282, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.090771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5108612060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17050, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.20089524537324904, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.12005701139569283, + "grad_norm": 0.0482548251748085, + "kept_token_count": 653.734375, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.990783554315567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00921645201742649, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.7376678466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0215545654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17060, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.1995638132095337, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.11799251884222031, + "grad_norm": 0.05521702021360397, + "kept_token_count": 654.928125, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0101, + "mgd_cos_sim": 0.9898997604846954, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010100230295211076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.88597717285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3598327636719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17070, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20062978118658065, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.12135576829314232, + "grad_norm": 0.051493119448423386, + "kept_token_count": 654.16875, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910336971282959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00896631134673953, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.83367614746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9731964111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20084450840950013, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.12268539369106293, + "grad_norm": 0.05895853787660599, + "kept_token_count": 653.8078125, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908263921737671, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009173609409481287, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.5854736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1324462890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17090, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.19993923008441924, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.1145862065255642, + "grad_norm": 0.06287116557359695, + "kept_token_count": 654.45625, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908261716365814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009173837397247553, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6140991210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.93178405761716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.19954072535037995, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.1366341196000576, + "grad_norm": 0.06348148733377457, + "kept_token_count": 654.9234375, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906947553157807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009305246826261281, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.85570373535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59608154296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17110, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.1994369938969612, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.11606923565268516, + "grad_norm": 0.04631439596414566, + "kept_token_count": 655.2, + "learning_rate": 1.274498147106221e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910087287425995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008991269301623106, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.52935485839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6005462646484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17120, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20020522475242614, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.1318697638809681, + "grad_norm": 0.0487055629491806, + "kept_token_count": 654.55, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906925201416016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009307475667446852, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.89671630859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8644622802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17130, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.19981821179389953, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.12268530428409577, + "grad_norm": 0.056381434202194214, + "kept_token_count": 654.4390625, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910071551799774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008992839138954878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5841857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.962353515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17140, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.20020677298307418, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.12314237356185913, + "grad_norm": 0.04457097128033638, + "kept_token_count": 654.171875, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909482419490814, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00905174035578966, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9440063476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4646850585938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17150, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.2001475840806961, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.11709224283695222, + "grad_norm": 0.04710168391466141, + "kept_token_count": 654.6, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912457704544068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008754231687635183, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.5829132080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3954132080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17160, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.20004057735204697, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.12581529170274736, + "grad_norm": 0.05145087093114853, + "kept_token_count": 654.4609375, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908831059932709, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009116891399025917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5898803710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9126281738281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17170, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19901950806379318, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.13491015657782554, + "grad_norm": 0.06199880689382553, + "kept_token_count": 655.4890625, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912790834903717, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00872092815116048, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.06553039550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8171722412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17180, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.20040675550699233, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.11435711160302162, + "grad_norm": 0.05855581909418106, + "kept_token_count": 654.2234375, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911841630935669, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008815843518823386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7851165771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8759765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17190, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.1984135091304779, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.13315766155719758, + "grad_norm": 0.07268859446048737, + "kept_token_count": 655.7109375, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905728757381439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009427128452807665, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3168426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89283447265626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17200, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20093365758657455, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.12236501500010491, + "grad_norm": 0.04727451503276825, + "kept_token_count": 653.9859375, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905554234981537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009444578550755978, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0284790039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.89751892089845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17210, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20010291337966918, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.14046143367886543, + "grad_norm": 0.0517997108399868, + "kept_token_count": 654.375, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909394562244416, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00906053576618433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1914001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8901062011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17220, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.1994967743754387, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.137334194034338, + "grad_norm": 0.059237442910671234, + "kept_token_count": 654.9484375, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.99053053855896, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009469455014914275, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2961853027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.27052917480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17230, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.19945110380649567, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.13285007402300836, + "grad_norm": 0.0662633404135704, + "kept_token_count": 655.428125, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912654161453247, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00873458432033658, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0507385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3203186035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17240, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.19946747720241548, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.12102366164326668, + "grad_norm": 0.044690728187561035, + "kept_token_count": 655.0828125, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911002695560456, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008899735566228629, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2751068115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4321624755859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17250, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.19895973801612854, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.12892695292830467, + "grad_norm": 0.05876404047012329, + "kept_token_count": 655.2765625, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907483339309693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009251641761511564, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5247375488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8054962158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17260, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.20024981647729873, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.12423275560140609, + "grad_norm": 0.05326443165540695, + "kept_token_count": 654.6234375, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904394030570984, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009560596384108066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4238677978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.19599609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17270, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.1998944789171219, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.12554678991436957, + "grad_norm": 0.04651352018117905, + "kept_token_count": 654.5890625, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.9917381942272187, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008261813595891, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2774169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1217468261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20075948238372804, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.11902767717838288, + "grad_norm": 0.0735640674829483, + "kept_token_count": 653.5546875, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9903988599777221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00960115110501647, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3578796386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4461700439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17290, + "valid_token_count": 817.721875 + }, + { + "actual_token_mask_ratio": 0.20059044361114503, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.11931923404335976, + "grad_norm": 0.04405559226870537, + "kept_token_count": 654.2859375, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912399649620056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008760035876184702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.97735595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2098907470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17300, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.19975152760744094, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.12573545947670936, + "grad_norm": 0.050661277025938034, + "kept_token_count": 654.9359375, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903371453285217, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009662858955562115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4394165039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5503295898437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17310, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.1990960031747818, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.13049837425351143, + "grad_norm": 0.046423979103565216, + "kept_token_count": 655.0921875, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908443093299866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009155677631497383, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9839660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.69456787109374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17320, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.199884994328022, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.1183390848338604, + "grad_norm": 0.04222813993692398, + "kept_token_count": 654.790625, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905989944934845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009400999546051026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1767883300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.56678466796876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17330, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.2005619153380394, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.12457163408398628, + "grad_norm": 0.06022017076611519, + "kept_token_count": 654.09375, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904492139816284, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009550789557397365, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.7610809326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59649963378905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17340, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19930365234613417, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.11876339092850685, + "grad_norm": 0.057937655597925186, + "kept_token_count": 655.1421875, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911275446414948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008872441854327917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.30393371582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.13422241210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20019286274909973, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.11553229093551635, + "grad_norm": 0.06597641855478287, + "kept_token_count": 654.5796875, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912212193012238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008778790198266506, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.23645935058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2209869384766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17360, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19861635118722915, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.1283361367881298, + "grad_norm": 0.05739235877990723, + "kept_token_count": 655.884375, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908323287963867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009167667012661695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.5588623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.99635009765626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17370, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20014150738716124, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.121172846108675, + "grad_norm": 0.06113301217556, + "kept_token_count": 654.6875, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905685484409332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009431448392570018, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6210510253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09718017578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17380, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.20055863857269288, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.12386210337281227, + "grad_norm": 0.06281117349863052, + "kept_token_count": 653.909375, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.991103857755661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008896144479513169, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0374359130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.07130737304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17390, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.20037703067064286, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.12983501702547073, + "grad_norm": 0.04342226684093475, + "kept_token_count": 654.0078125, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908589005470276, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00914110578596592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7420623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.69044494628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17400, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.19938840121030807, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.1228187121450901, + "grad_norm": 0.06572292000055313, + "kept_token_count": 655.096875, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905268847942352, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009473113156855107, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.86719665527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4233856201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.1998842477798462, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.12302765399217605, + "grad_norm": 0.07758008688688278, + "kept_token_count": 654.4921875, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900454938411712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009954496752470732, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9641876220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.08494262695314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17420, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.2002212166786194, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.12468942552804947, + "grad_norm": 0.049906693398952484, + "kept_token_count": 654.6390625, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910118579864502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008988127764314412, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.15960388183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.31878356933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17430, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.19999469369649886, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.1198967844247818, + "grad_norm": 0.07203864306211472, + "kept_token_count": 654.7265625, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907869279384613, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00921307373791933, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.67576904296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5421508789062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17440, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20010351091623307, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.11601819396018982, + "grad_norm": 0.050279222428798676, + "kept_token_count": 654.4390625, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.9916830003261566, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00831700637936592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5842346191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.09636840820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17450, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19950585514307023, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.12392102032899857, + "grad_norm": 0.06782670319080353, + "kept_token_count": 654.86875, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906104862689972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009389505349099636, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.02362365722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.88165893554685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17460, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.20029575675725936, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.12538013756275176, + "grad_norm": 0.0719446986913681, + "kept_token_count": 654.5265625, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0099, + "mgd_cos_sim": 0.9900589108467102, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009941092878580093, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.05065002441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32867126464845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17470, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.20018958300352097, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.1258188769221306, + "grad_norm": 0.04726269841194153, + "kept_token_count": 654.3140625, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909304022789002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009069605637341738, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.04640502929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.87842712402346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17480, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20006830841302872, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.12207123711705208, + "grad_norm": 0.047104619443416595, + "kept_token_count": 654.4859375, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905179679393769, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009482034761458636, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.043017578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4004211425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17490, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.20088194012641908, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.12106670215725898, + "grad_norm": 0.04110107198357582, + "kept_token_count": 653.8046875, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.99150829911232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008491695020347833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0880126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9844909667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17500, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20015052258968352, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.12672139331698418, + "grad_norm": 0.05299760401248932, + "kept_token_count": 654.125, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.9917199492454529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008280033897608518, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.07688598632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2266876220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17510, + "valid_token_count": 817.8109375 + }, + { + "actual_token_mask_ratio": 0.2007821187376976, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.1308411441743374, + "grad_norm": 0.058302026242017746, + "kept_token_count": 654.271875, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0103, + "mgd_cos_sim": 0.9897040486335754, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010295937396585941, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7290466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3526641845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17520, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.19983742386102676, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.12786128222942353, + "grad_norm": 0.03997068479657173, + "kept_token_count": 654.646875, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911942601203918, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008805753011256457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2976806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8800079345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17530, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.1998911529779434, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.1276371531188488, + "grad_norm": 0.06828579306602478, + "kept_token_count": 654.7046875, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9906429886817932, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009357009269297122, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3984802246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.23976135253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20013324320316314, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.12378569915890694, + "grad_norm": 0.058817557990550995, + "kept_token_count": 654.14375, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908957302570343, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009104271419346333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6842102050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.78684387207034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17550, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.1992647036910057, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.1275322012603283, + "grad_norm": 0.056874584406614304, + "kept_token_count": 655.05, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906517684459686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009348239563405514, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.31861267089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3492065429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17560, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.2010713815689087, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.11648810058832168, + "grad_norm": 0.04868118464946747, + "kept_token_count": 653.9953125, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911449193954468, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00885510416701436, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1020904541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.94237060546874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17570, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.20154687911272048, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.1293371506035328, + "grad_norm": 0.04149453714489937, + "kept_token_count": 653.065625, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908910155296325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00910898419097066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9786804199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.43174133300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17580, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.19988818913698198, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.12199554517865181, + "grad_norm": 0.06087857857346535, + "kept_token_count": 654.6546875, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909332811832428, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009066712018102408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.36683349609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.53651733398436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17590, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19950266927480698, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.12202566415071488, + "grad_norm": 0.05879378318786621, + "kept_token_count": 654.90625, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910988092422486, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00890119131654501, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.38388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9584259033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17600, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.20032970756292343, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.12071654498577118, + "grad_norm": 0.04996786639094353, + "kept_token_count": 654.4578125, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912110507488251, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008788935746997595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.54371643066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.90656127929685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17610, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20102520883083344, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.11992216929793358, + "grad_norm": 0.05644307658076286, + "kept_token_count": 653.8859375, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906867802143097, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00931320758536458, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.03241271972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3208465576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19964303970336914, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.12588780820369722, + "grad_norm": 0.05985206738114357, + "kept_token_count": 655.13125, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9902887403964996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00971125615760684, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.11328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.02161865234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17630, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19954263865947724, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.12571757584810256, + "grad_norm": 0.05500595271587372, + "kept_token_count": 655.0984375, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907807648181916, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009219235368072986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.35117797851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1916809082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17640, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.19901255518198013, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.12347690537571906, + "grad_norm": 0.04782296344637871, + "kept_token_count": 655.39375, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909903585910798, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009009635169059038, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.59693298339846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0901763916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17650, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19924114942550658, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.11842036247253418, + "grad_norm": 0.056962914764881134, + "kept_token_count": 655.5359375, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901524722576142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009847528208047152, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3555480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6752532958984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17660, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.19938068687915803, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.11647284775972366, + "grad_norm": 0.06463773548603058, + "kept_token_count": 655.10625, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904060125350952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00959397740662098, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.81218872070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.30637512207034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17670, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2007270947098732, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.12605826258659364, + "grad_norm": 0.04245121404528618, + "kept_token_count": 653.8796875, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905013859272003, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009498622734099627, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.944482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.23528747558595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17680, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19950920790433885, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.1345967411994934, + "grad_norm": 0.054490525275468826, + "kept_token_count": 655.2046875, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910154163837432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0089845965616405, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6710906982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7959808349609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17690, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.2003455266356468, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.1255228638648987, + "grad_norm": 0.053320735692977905, + "kept_token_count": 654.2484375, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.990827476978302, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009172517992556094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5804443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.95691833496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17700, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19939668476581573, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.1244935393333435, + "grad_norm": 0.05815953388810158, + "kept_token_count": 654.946875, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.9905636966228485, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00943629713729024, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5746124267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4362518310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17710, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.1999580219388008, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.11971236690878868, + "grad_norm": 0.06080963835120201, + "kept_token_count": 654.8453125, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908538162708282, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009146186895668507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.61049194335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.54539489746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17720, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20088883489370346, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.13486869111657143, + "grad_norm": 0.04418634995818138, + "kept_token_count": 654.0265625, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907684683799743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009231521654874087, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7465362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49901428222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17730, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.1999901384115219, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.10936015695333481, + "grad_norm": 0.05616103857755661, + "kept_token_count": 654.6265625, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.009, + "mgd_cos_sim": 0.990990149974823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009009845089167356, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.4878814697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3417602539063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17740, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.2003154918551445, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.1206228829920292, + "grad_norm": 0.05403450131416321, + "kept_token_count": 654.5796875, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913733661174774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008626632019877433, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.5141632080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6842987060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17750, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19960627406835557, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.11419039964675903, + "grad_norm": 0.051190249621868134, + "kept_token_count": 655.025, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909524857997895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009047509264200925, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.5668426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1492065429687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17760, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19976161271333695, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.12177718579769134, + "grad_norm": 0.058837346732616425, + "kept_token_count": 654.784375, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910260617733002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008973935805261134, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3584350585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.31833801269534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17770, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.1992090091109276, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.12369438335299492, + "grad_norm": 0.08934129029512405, + "kept_token_count": 655.3421875, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909982919692993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00900170747190714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.69849548339846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3304107666016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17780, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.19957281947135924, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.12394452691078187, + "grad_norm": 0.05907556414604187, + "kept_token_count": 654.86875, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911129236221313, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008887077309191227, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8192169189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.535986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17790, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2004684180021286, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.11953557953238488, + "grad_norm": 0.06549867242574692, + "kept_token_count": 654.3734375, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907176971435547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009282297920435667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.94329223632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3570831298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17800, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.20001716166734695, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.12895879969000817, + "grad_norm": 0.05080156400799751, + "kept_token_count": 654.578125, + "learning_rate": 1.162466013295864e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910388708114624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008961119130253792, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2676300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4529144287109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20031387507915496, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.1297013431787491, + "grad_norm": 0.05606063827872276, + "kept_token_count": 654.125, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908277511596679, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009172255452722312, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.09307250976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7152465820312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17820, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.19984347075223924, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.12292011454701424, + "grad_norm": 0.06701008975505829, + "kept_token_count": 654.7265625, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911618292331695, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00883817058056593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.8207122802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8285430908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17830, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20001459121704102, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.1332777552306652, + "grad_norm": 0.06132035702466965, + "kept_token_count": 654.6140625, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0102, + "mgd_cos_sim": 0.9897833406925202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.010216650739312171, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.71214904785154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4233123779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17840, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.20079819113016129, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.12708517760038376, + "grad_norm": 0.061526838690042496, + "kept_token_count": 653.6875, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909042894840241, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009095708187669516, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2454833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.86724853515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17850, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.19982847422361374, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.12371591478586197, + "grad_norm": 0.04169169068336487, + "kept_token_count": 654.925, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911669671535492, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008833034057170152, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6228271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5195251464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17860, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.1995606780052185, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.13011092618107795, + "grad_norm": 0.04926476255059242, + "kept_token_count": 655.1578125, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912609398365021, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008739053271710873, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.25604248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9032348632812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17870, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.1997364044189453, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.11575916185975074, + "grad_norm": 0.06972280889749527, + "kept_token_count": 654.73125, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908996939659118, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009100309573113918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.43720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1091339111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17880, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.19936235547065734, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.12271040156483651, + "grad_norm": 0.04337219148874283, + "kept_token_count": 655.0421875, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907535135746002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009246499929577112, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.81883239746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7930389404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17890, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19982516020536423, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.12060473710298539, + "grad_norm": 0.04333655908703804, + "kept_token_count": 654.9375, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.9914599776268005, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00854001808911562, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7558624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.12818298339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17900, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.20022032856941224, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.11968428045511245, + "grad_norm": 0.06954730302095413, + "kept_token_count": 654.5546875, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909449458122254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009055049903690816, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4922302246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6964508056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17910, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20173091888427735, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.12248405739665032, + "grad_norm": 0.05436502769589424, + "kept_token_count": 652.81875, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908914566040039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00910853100940585, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.981787109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3599609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17920, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.20009544789791106, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.1182814247906208, + "grad_norm": 0.04880101606249809, + "kept_token_count": 654.4546875, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911890089511871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008810992632061244, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7955017089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.93181762695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17930, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.20059193074703216, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.12924168780446052, + "grad_norm": 0.0804559588432312, + "kept_token_count": 654.040625, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.990586805343628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009413186367601157, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.38612365722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7303985595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17940, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19963843673467635, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.12535079345107078, + "grad_norm": 0.05802726373076439, + "kept_token_count": 654.8203125, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910957753658295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00890422984957695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5168731689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3879699707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17950, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.19956225156784058, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.12407362312078477, + "grad_norm": 0.06447485834360123, + "kept_token_count": 655.0328125, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0097, + "mgd_cos_sim": 0.9903368771076202, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009663113299757243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5884704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6841156005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17960, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.20052401274442672, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.12662901282310485, + "grad_norm": 0.10226797312498093, + "kept_token_count": 654.1015625, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904339134693145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009566089324653148, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5575439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1345916748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1991923451423645, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.11685300096869469, + "grad_norm": 0.053080108016729355, + "kept_token_count": 655.340625, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9908323287963867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009167674463242292, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8509063720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7453369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17980, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20049446523189546, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.12202187180519104, + "grad_norm": 0.10977630317211151, + "kept_token_count": 654.340625, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910243272781372, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008975668903440237, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.56752014160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0137420654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17990, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20031839460134507, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.12018076926469803, + "grad_norm": 0.05390943959355354, + "kept_token_count": 654.6875, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9913280367851257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008671945612877607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7065002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.10451354980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18000, + "valid_token_count": 818.6828125 + }, + { + "actual_token_mask_ratio": 0.20060093998908995, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.11929340139031411, + "grad_norm": 0.04171936959028244, + "kept_token_count": 654.0625, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912026345729827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008797371108084918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8857940673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11570434570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19987326860427856, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.12702143117785453, + "grad_norm": 0.04817148298025131, + "kept_token_count": 654.7703125, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911303460597992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00886966185644269, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3303924560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6593322753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18020, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.1999111667275429, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.1231396846473217, + "grad_norm": 0.05297398194670677, + "kept_token_count": 654.7703125, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909056425094604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009094351530075073, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.33519287109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3320343017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18030, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.19971963316202163, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.13227496147155762, + "grad_norm": 0.049307554960250854, + "kept_token_count": 654.70625, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.9915144443511963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008485548105090857, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7064208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.81199951171874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18040, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2006598487496376, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.13433672860264778, + "grad_norm": 0.04756542667746544, + "kept_token_count": 654.0734375, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.991669374704361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008330631349235774, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.22218017578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.62493591308595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18050, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20034806281328202, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.12258114367723465, + "grad_norm": 0.05809124559164047, + "kept_token_count": 654.559375, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907703280448914, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009229661989957094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2930480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6099395751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18060, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.20026667714118956, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.12769771441817285, + "grad_norm": 0.040269289165735245, + "kept_token_count": 654.4484375, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911345958709716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008865405712276698, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1540161132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.10760803222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18070, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19999667555093764, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.12491377592086791, + "grad_norm": 0.04689788818359375, + "kept_token_count": 654.6, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.990630692243576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009369302913546563, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4274444580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2561767578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18080, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.19937657415866852, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.12644738033413888, + "grad_norm": 0.04927777498960495, + "kept_token_count": 655.1875, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909745812416076, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009025415219366551, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7543609619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.66796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18090, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.19922143518924712, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.11369165778160095, + "grad_norm": 0.046747446060180664, + "kept_token_count": 655.1796875, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907290995121002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009270893782377243, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1362335205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.98500061035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18100, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.20054664313793183, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.12688101977109909, + "grad_norm": 0.04743083938956261, + "kept_token_count": 654.075, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990865820646286, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009134156629443169, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6464538574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3194641113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18110, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.1993236944079399, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.12118356674909592, + "grad_norm": 0.06445840746164322, + "kept_token_count": 655.234375, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908903062343597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00910968827083707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2887939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09503479003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.20050622820854186, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.12453912049531937, + "grad_norm": 0.0669727474451065, + "kept_token_count": 654.340625, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9902169704437256, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009783040918409824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.37762756347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3631958007812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18130, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19978240728378296, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.1272997036576271, + "grad_norm": 0.04625562205910683, + "kept_token_count": 654.7703125, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.9914827764034271, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008517233561724425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.93040466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0126159667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18140, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20044819116592408, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.12145477905869484, + "grad_norm": 0.06248428300023079, + "kept_token_count": 654.1265625, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.9917312741279602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008268733229488134, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.83916625976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8115539550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18150, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20064421445131303, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.12477261871099472, + "grad_norm": 0.05343754217028618, + "kept_token_count": 653.8828125, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905168533325195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009483137167990208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.29454040527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2522338867187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18160, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.19904284179210663, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.12404333502054214, + "grad_norm": 0.04342406243085861, + "kept_token_count": 655.3828125, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913614451885223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008638541866093873, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9865417480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1116668701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18170, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19848520010709764, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.12270049676299095, + "grad_norm": 0.044006720185279846, + "kept_token_count": 655.778125, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909629404544831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00903705544769764, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.643408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1256988525391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18180, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19998832494020463, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.1297477051615715, + "grad_norm": 0.053270190954208374, + "kept_token_count": 654.371875, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912840902805329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008715905249118805, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.25316467285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.64272155761716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18190, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.19967252910137176, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.12038243114948273, + "grad_norm": 0.057338301092386246, + "kept_token_count": 655.1703125, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911155700683594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008884431421756744, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.62757568359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6547119140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18200, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20001693814992905, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.12251024767756462, + "grad_norm": 0.06802491098642349, + "kept_token_count": 654.5953125, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907908320426941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009209157153964042, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5374420166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.18287353515626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18210, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20083554983139038, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.13093086183071137, + "grad_norm": 0.05891226977109909, + "kept_token_count": 654.0625, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990902453660965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009097554255276918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8130676269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1764251708984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18220, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20054255276918412, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.11751044616103172, + "grad_norm": 0.05750951170921326, + "kept_token_count": 653.9015625, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907576620578766, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009242336545139552, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.06732177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4609008789063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18230, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.20127155035734176, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.1330620288848877, + "grad_norm": 0.05508100241422653, + "kept_token_count": 653.33125, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909753561019897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009024644549936057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0333953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7848419189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18240, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.1998351365327835, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.12477931082248687, + "grad_norm": 0.051461361348629, + "kept_token_count": 654.5328125, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912773966789246, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008722591120749712, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9569946289063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.06793823242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.19990454614162445, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.12444661185145378, + "grad_norm": 0.04461727663874626, + "kept_token_count": 654.646875, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.990853899717331, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009146097954362631, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6617065429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7296600341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18260, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19977940917015075, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.13286222666501998, + "grad_norm": 0.05161620303988457, + "kept_token_count": 654.790625, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904652655124664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009534714743494987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6578002929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3544860839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18270, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19952931106090546, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.11452636122703552, + "grad_norm": 0.045239344239234924, + "kept_token_count": 655.0390625, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913824737071991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00861753523349762, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.354052734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9430694580078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18280, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2003740757703781, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.1174422025680542, + "grad_norm": 0.048324521631002426, + "kept_token_count": 654.06875, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9908700406551361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009129960183054208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8847930908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.68614501953124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18290, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.20092677026987077, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.12048199921846389, + "grad_norm": 0.06195906549692154, + "kept_token_count": 653.771875, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.991284704208374, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008715278934687376, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.62445678710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3354125976563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18300, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19999805837869644, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.1204767219722271, + "grad_norm": 0.050711557269096375, + "kept_token_count": 654.628125, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907460987567902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009253901988267898, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7720031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9587768554687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20031650960445405, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.1190066508948803, + "grad_norm": 0.05045389011502266, + "kept_token_count": 654.3703125, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0096, + "mgd_cos_sim": 0.9904049575328827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009595031570643186, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1172393798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2813690185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18320, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2000440016388893, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.13059799149632453, + "grad_norm": 0.05089013651013374, + "kept_token_count": 654.453125, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9905083835124969, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009491609036922454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0826354980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94329528808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18330, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20017785727977752, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.1281910091638565, + "grad_norm": 0.0630015879869461, + "kept_token_count": 654.6375, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0098, + "mgd_cos_sim": 0.9901815176010131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009818490967154503, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.01563720703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.62616271972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18340, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.19970948994159698, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.12191934213042259, + "grad_norm": 0.04482425004243851, + "kept_token_count": 654.83125, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.991474324464798, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008525673858821392, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2908233642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.23187255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18350, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.1997888997197151, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.12008895203471184, + "grad_norm": 0.07401730865240097, + "kept_token_count": 654.8578125, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9909527182579041, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00904728127643466, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8798522949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.63017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18360, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1990613341331482, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.1284518599510193, + "grad_norm": 0.04959679767489433, + "kept_token_count": 655.315625, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911375403404236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00886247269809246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4148895263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1078369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18370, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19970993399620057, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.1286646917462349, + "grad_norm": 0.05904141813516617, + "kept_token_count": 654.946875, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911436855792999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008856326434761285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9513214111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.95240478515626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18380, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.1991122081875801, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.12619126364588737, + "grad_norm": 0.05304032191634178, + "kept_token_count": 655.5796875, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912669777870178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008733015414327383, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3667297363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.359716796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18390, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.19969195872545242, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.13290934413671493, + "grad_norm": 0.05503743514418602, + "kept_token_count": 654.940625, + "learning_rate": 1.068193811655972e-05, + "loss": 0.009, + "mgd_cos_sim": 0.991024786233902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008975216560065747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3496612548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6361083984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18400, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.1997306764125824, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.12568895518779755, + "grad_norm": 0.05619596689939499, + "kept_token_count": 655.2625, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909354984760285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00906450841575861, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1664764404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7064910888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18410, + "valid_token_count": 818.8 + }, + { + "actual_token_mask_ratio": 0.20081350207328796, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.11906293109059334, + "grad_norm": 0.049128662794828415, + "kept_token_count": 653.9703125, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0082, + "mgd_cos_sim": 0.991840523481369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008159476798027754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.51866455078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.35144958496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18420, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20098249167203902, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.1289438560605049, + "grad_norm": 0.04767739400267601, + "kept_token_count": 653.8953125, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910530030727387, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008947013691067696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.58655700683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9102386474609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18430, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.2007371351122856, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.11802590042352676, + "grad_norm": 0.05265139788389206, + "kept_token_count": 654.15625, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0095, + "mgd_cos_sim": 0.9904985845088958, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009501404222100973, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.80669860839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.13067016601565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18440, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19922886788845062, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.12249101772904396, + "grad_norm": 0.04850434511899948, + "kept_token_count": 655.3171875, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.009, + "mgd_cos_sim": 0.99095379114151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009046216681599616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8299926757812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.77686462402346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18450, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2000729814171791, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.12772158160805702, + "grad_norm": 0.04938919097185135, + "kept_token_count": 654.53125, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.9914636015892029, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008536399994045495, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5283630371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0824951171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18460, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20090076923370362, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.11631428524851799, + "grad_norm": 0.04722332954406738, + "kept_token_count": 653.884375, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910175502300262, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008982465229928493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8094757080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1809417724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18470, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20085036456584932, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.11895168125629425, + "grad_norm": 0.04634212702512741, + "kept_token_count": 654.1453125, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9912981569766999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008701856900006532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.62059020996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.10630493164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18480, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.20003560334444045, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.12343852743506431, + "grad_norm": 0.04618227854371071, + "kept_token_count": 654.5265625, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.990652596950531, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00934740211814642, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1010437011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3374328613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19851405620574952, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.11544430255889893, + "grad_norm": 0.042314112186431885, + "kept_token_count": 655.90625, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0084, + "mgd_cos_sim": 0.9916462302207947, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00835376065224409, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0550140380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.6038452148438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19959687441587448, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.12841158285737037, + "grad_norm": 0.04937959462404251, + "kept_token_count": 655.121875, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910580933094024, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008941905666142702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.83848266601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6646759033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18510, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19994766414165496, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.1150453045964241, + "grad_norm": 0.059900470077991486, + "kept_token_count": 654.9484375, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912172377109527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008782764803618193, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7986724853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.45392456054685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18520, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.20072561502456665, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.12166940048336983, + "grad_norm": 0.06057126447558403, + "kept_token_count": 654.078125, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0085, + "mgd_cos_sim": 0.9915367126464844, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0084632633253932, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5061859130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0330322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18530, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.1995282456278801, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.12387429550290108, + "grad_norm": 0.044867243617773056, + "kept_token_count": 655.1703125, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.009, + "mgd_cos_sim": 0.990973049402237, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00902694258838892, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.33824462890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.94655151367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18540, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20095837265253066, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.12718903869390488, + "grad_norm": 0.06283413618803024, + "kept_token_count": 653.840625, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913772106170654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008622806519269943, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2354766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.70127868652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18550, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.20031366497278214, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.1252107471227646, + "grad_norm": 0.12758710980415344, + "kept_token_count": 654.15, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910331964492798, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008966803550720215, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8348114013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18560, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.1994238018989563, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.12231578379869461, + "grad_norm": 0.05837077647447586, + "kept_token_count": 654.728125, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906593084335327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009340687189251184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.7100372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6949401855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18570, + "valid_token_count": 817.8171875 + }, + { + "actual_token_mask_ratio": 0.20050710141658784, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.11337251216173172, + "grad_norm": 0.05238906666636467, + "kept_token_count": 654.4140625, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906509697437287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009349042549729347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.48363342285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.15330200195314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18580, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.200624780356884, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.12157348468899727, + "grad_norm": 0.05868643894791603, + "kept_token_count": 654.2125, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910538077354432, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008946198411285878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.03143615722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.105615234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18590, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.199793803691864, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.12000896483659744, + "grad_norm": 0.04915480315685272, + "kept_token_count": 654.5875, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0084, + "mgd_cos_sim": 0.9915914833545685, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408528659492731, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.92034301757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7937438964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18600, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.20001306086778642, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.11855055764317513, + "grad_norm": 0.05055814981460571, + "kept_token_count": 654.815625, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.991383409500122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008616588171571493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7529815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7640441894531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18610, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.20155440121889115, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.1170965202152729, + "grad_norm": 0.0466250479221344, + "kept_token_count": 653.146875, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911708116531373, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008829191047698259, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.0149383544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89320068359376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18620, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.19890818893909454, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.11892541125416756, + "grad_norm": 0.04239002615213394, + "kept_token_count": 655.6890625, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9910897433757782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00891025671735406, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9483123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.46416931152345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18630, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.19980716854333877, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.1171333335340023, + "grad_norm": 0.05468851327896118, + "kept_token_count": 655.028125, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0092, + "mgd_cos_sim": 0.9907942652702332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009205716010183097, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.54356079101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.17193603515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18640, + "valid_token_count": 818.5875 + }, + { + "actual_token_mask_ratio": 0.200051748752594, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.11918900385499001, + "grad_norm": 0.0680711641907692, + "kept_token_count": 654.5875, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911614060401917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008838591445237399, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.80438537597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.98334045410155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.19948683828115463, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.12084714025259018, + "grad_norm": 0.04824730008840561, + "kept_token_count": 655.1421875, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913862884044647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008613707963377237, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0445587158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9185791015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18660, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.1999274343252182, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.11699943616986275, + "grad_norm": 0.06466619670391083, + "kept_token_count": 654.6359375, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909144163131713, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009085595142096281, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.50115356445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0335174560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18670, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19971734583377837, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.12112370058894158, + "grad_norm": 0.05586666613817215, + "kept_token_count": 655.1859375, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913577020168305, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008642293978482485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.59703674316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2272247314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18680, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.20003228336572648, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.1237249381840229, + "grad_norm": 0.04512105509638786, + "kept_token_count": 654.328125, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0087, + "mgd_cos_sim": 0.9913327276706696, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008667274098843335, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.62096252441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1366424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18690, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20035809725522996, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.12169757038354874, + "grad_norm": 0.05468776449561119, + "kept_token_count": 654.2484375, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9911892831325531, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008810704201459884, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8548889160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.32855834960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18700, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.20068961679935454, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.11630397215485573, + "grad_norm": 0.1423678696155548, + "kept_token_count": 654.1515625, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9906836092472077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009316390659660102, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.71527404785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.32623291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18710, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19928289949893951, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.12459937408566475, + "grad_norm": 0.06659697741270065, + "kept_token_count": 654.9859375, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.009, + "mgd_cos_sim": 0.9910433053970337, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008956685196608305, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.98203735351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.239208984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18720, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.20000613629817962, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.12257522344589233, + "grad_norm": 0.05248090624809265, + "kept_token_count": 654.775, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.01, + "mgd_cos_sim": 0.9900365054607392, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00996349873021245, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2111541748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9161437988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18730, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.2002914071083069, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.13256537020206452, + "grad_norm": 0.055694643408060074, + "kept_token_count": 654.8, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0089, + "mgd_cos_sim": 0.9911218643188476, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008878133539110423, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9628845214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0878112792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18740, + "valid_token_count": 818.8015625 + }, + { + "actual_token_mask_ratio": 0.20024783313274383, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.1268270120024681, + "grad_norm": 0.03931112959980965, + "kept_token_count": 654.4421875, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0082, + "mgd_cos_sim": 0.9917782127857209, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008221774362027645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0974395751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3063201904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18750, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.20062924772500992, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.12125028669834137, + "grad_norm": 0.06309893727302551, + "kept_token_count": 654.153125, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0088, + "mgd_cos_sim": 0.9912195444107056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008780456986278296, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.41400451660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.62254638671874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18760, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.1996988132596016, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.12340262681245803, + "grad_norm": 0.062268100678920746, + "kept_token_count": 654.9453125, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0082, + "mgd_cos_sim": 0.9918064773082733, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00819351151585579, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2578430175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4460510253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18770, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20058621615171432, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.1176472246646881, + "grad_norm": 0.04027304798364639, + "kept_token_count": 654.2234375, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0086, + "mgd_cos_sim": 0.9913597702980042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008640226069837808, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.93031311035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1501220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18780, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.19935495853424073, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.13075477331876756, + "grad_norm": 0.04686214029788971, + "kept_token_count": 655.025, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0093, + "mgd_cos_sim": 0.9907046258449554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009295360837131738, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.61206970214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.92809143066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20040716379880905, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.13411233350634574, + "grad_norm": 0.05070391297340393, + "kept_token_count": 654.64375, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0094, + "mgd_cos_sim": 0.990566486120224, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009433514159172773, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.10475463867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3033935546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18800, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.1998980462551117, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.12144949436187744, + "grad_norm": 0.06421314179897308, + "kept_token_count": 654.7671875, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0091, + "mgd_cos_sim": 0.9909343957901001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009065606258809566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.97374267578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.26954345703126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18810, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19972752928733825, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.12609679847955704, + "grad_norm": 0.03639940172433853, + "kept_token_count": 655.0609375, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0084, + "mgd_cos_sim": 0.9915782153606415, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008421785011887551, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3874572753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.11304626464846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18820, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.20000389069318772, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.12786667197942733, + "grad_norm": 0.056524693965911865, + "kept_token_count": 654.4734375, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0083, + "mgd_cos_sim": 0.9916723608970642, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0083276329562068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6659393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1130676269531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18830, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20053455978631973, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.12797383666038514, + "grad_norm": 0.047251246869564056, + "kept_token_count": 654.153125, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9912394165992737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00876058042049408, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.21729431152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0017883300781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18840, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.19815240651369095, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.12560559511184693, + "grad_norm": 0.05644700303673744, + "kept_token_count": 656.2546875, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911312818527221, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00886871013790369, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9101043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2687927246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18850, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.20010706931352615, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.12432088553905488, + "grad_norm": 0.04717004671692848, + "kept_token_count": 654.334375, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911499679088592, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008850038424134255, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.64060974121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4032867431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18860, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20081621557474136, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.13051779791712761, + "grad_norm": 0.05105507746338844, + "kept_token_count": 654.0359375, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916721880435944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008327810000628233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9522674560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.15785217285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18870, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.1996924489736557, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.11991727277636528, + "grad_norm": 0.05526730790734291, + "kept_token_count": 654.9765625, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914068520069123, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008593151997774839, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4247650146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3808959960937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18880, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20032074600458144, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.126093028485775, + "grad_norm": 0.04287660866975784, + "kept_token_count": 654.134375, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918324589729309, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008167540840804576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.20843811035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.58369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18890, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.199997179210186, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.12741166353225708, + "grad_norm": 0.051333677023649216, + "kept_token_count": 654.659375, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.990884131193161, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009115881100296975, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5345703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18900, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.200883749127388, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.11616251021623611, + "grad_norm": 0.0464138500392437, + "kept_token_count": 653.8734375, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991355550289154, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008644449803978205, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5133361816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4802001953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18910, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2004695102572441, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.11975313946604729, + "grad_norm": 0.04581386223435402, + "kept_token_count": 654.2984375, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911644041538239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008835584297776223, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.39028015136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9233154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18920, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20136540085077287, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.11838965192437172, + "grad_norm": 0.060601163655519485, + "kept_token_count": 653.4734375, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9907085955142975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009291390422731638, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.839453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.602294921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18930, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20105957686901094, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.1170403465628624, + "grad_norm": 0.08630188554525375, + "kept_token_count": 653.815625, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9907266676425934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0092733315192163, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.05393981933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.94533081054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18940, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.1998900443315506, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.12008137851953507, + "grad_norm": 0.054816972464323044, + "kept_token_count": 654.73125, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.990749454498291, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009250559378415346, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7622314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1535949707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18950, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20048856884241104, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.12382183820009232, + "grad_norm": 0.058981381356716156, + "kept_token_count": 654.18125, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9907425582408905, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00925744092091918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.92950744628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5633239746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18960, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.19978518784046173, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.1258055843412876, + "grad_norm": 0.04080919176340103, + "kept_token_count": 654.5484375, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912754774093628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008724518958479166, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0271514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.13011474609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18970, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19993919432163237, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.11862456947565078, + "grad_norm": 0.14000466465950012, + "kept_token_count": 654.659375, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.991207504272461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008792479895055294, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.89133605957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.95989685058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18980, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2003063976764679, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.12121899351477623, + "grad_norm": 0.06609427183866501, + "kept_token_count": 654.5203125, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.991078895330429, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008921104948967695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.18348693847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8349884033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18990, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19935639947652817, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.12570198625326157, + "grad_norm": 0.0697547048330307, + "kept_token_count": 655.1765625, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0096, + "mgd_cos_sim": 0.9903812348842621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009618756547570228, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.47357788085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1411102294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19000, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.1996323585510254, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.126782888174057, + "grad_norm": 0.05112775042653084, + "kept_token_count": 654.9046875, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915033161640168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00849667815491557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4443664550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.485986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19010, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19974423199892044, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.12101153060793876, + "grad_norm": 0.05634378641843796, + "kept_token_count": 654.671875, + "learning_rate": 9.711026860773595e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9909968793392181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009003121871501207, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5134704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11777648925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19020, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.2007444307208061, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.12405232042074203, + "grad_norm": 0.06724805384874344, + "kept_token_count": 654.0765625, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913164377212524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00868357066065073, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3469177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41530456542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19030, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19879308342933655, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.13882860094308852, + "grad_norm": 0.04429874196648598, + "kept_token_count": 655.8859375, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.991328752040863, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008671231660991907, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.62293395996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.01991271972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19040, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.20067086368799208, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.12873764634132384, + "grad_norm": 0.047503747045993805, + "kept_token_count": 653.8984375, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9909031450748443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009096860326826573, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.557666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6237335205078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19050, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20034826993942262, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.12805611565709113, + "grad_norm": 0.0564100407063961, + "kept_token_count": 654.265625, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.991154134273529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008845864143222571, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.695458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1468170166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19060, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2006176769733429, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.12699616849422454, + "grad_norm": 0.04586295783519745, + "kept_token_count": 654.24375, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9908980190753937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009101989772170782, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4604461669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2309539794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19070, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.2003592237830162, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.12301621660590172, + "grad_norm": 0.04751722142100334, + "kept_token_count": 654.31875, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9907691955566407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009230807982385158, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.232763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3754364013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19080, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2001348078250885, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.12727853953838347, + "grad_norm": 0.042461197823286057, + "kept_token_count": 654.365625, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9909433782100677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009056637249886989, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.19773559570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.19122314453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19090, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19910425394773484, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.12142090275883674, + "grad_norm": 0.07969643920660019, + "kept_token_count": 655.3625, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9912099123001099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008790088910609483, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0144775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.35374755859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19100, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.199891497194767, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.13733526468276977, + "grad_norm": 0.043395817279815674, + "kept_token_count": 654.9375, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922508776187897, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007749138120561838, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2338439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7228149414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19110, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.20048351287841798, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.1256660759449005, + "grad_norm": 0.05338054150342941, + "kept_token_count": 653.921875, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911581039428711, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008841892518103123, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1841766357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.79896850585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19120, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.19942634105682372, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.130864030867815, + "grad_norm": 0.04898259416222572, + "kept_token_count": 655.3984375, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9912360489368439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008763962797820568, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.42945556640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2447082519531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19130, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.19954265803098678, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.1259574241936207, + "grad_norm": 0.040185365825891495, + "kept_token_count": 654.7828125, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0094, + "mgd_cos_sim": 0.9906220257282257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009377974178642034, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8500732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6475463867188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19140, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.20109448879957198, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.12302915453910827, + "grad_norm": 0.053469471633434296, + "kept_token_count": 653.6796875, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913017272949218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00869827326387167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.9900329589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7987457275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19150, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19976974427700042, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.13281987607479095, + "grad_norm": 0.04839104413986206, + "kept_token_count": 654.7546875, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0094, + "mgd_cos_sim": 0.9906476616859436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009352332912385464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.23460083007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9365295410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19160, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19922865629196168, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.11644840613007545, + "grad_norm": 0.05907147005200386, + "kept_token_count": 655.5859375, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914380252361298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008561976253986359, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.28990173339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.40423278808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19170, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.19973478764295577, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.12623933777213098, + "grad_norm": 0.04093946889042854, + "kept_token_count": 654.9234375, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917685270309449, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008231477160006762, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.30889892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2316589355469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19180, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20065180212259293, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.12423608973622322, + "grad_norm": 0.04882688820362091, + "kept_token_count": 654.2484375, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911963522434235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00880364840850234, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0017333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2380096435547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19190, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20059871971607207, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.13644981309771537, + "grad_norm": 0.06292178481817245, + "kept_token_count": 654.0140625, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915912389755249, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408765587955714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1099334716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3046813964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19200, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.20105060935020447, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.1257594034075737, + "grad_norm": 0.051474593579769135, + "kept_token_count": 653.5015625, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914488852024078, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008551111817359925, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7832733154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9819305419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19210, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.19982366114854813, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.11875389814376831, + "grad_norm": 0.04542588070034981, + "kept_token_count": 654.8671875, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914273500442505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008572661224752665, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.48023376464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2187164306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19220, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19899354130029678, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.13083931654691697, + "grad_norm": 0.06361424922943115, + "kept_token_count": 655.0953125, + "learning_rate": 9.387545045139381e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910042643547058, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008995732478797435, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1512817382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.19371643066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19230, + "valid_token_count": 817.8375 + }, + { + "actual_token_mask_ratio": 0.19999206364154815, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.12435532286763192, + "grad_norm": 0.055673662573099136, + "kept_token_count": 654.40625, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917610049247741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008238997496664523, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4821441650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8829681396484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19240, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.19979656636714935, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.1364349476993084, + "grad_norm": 0.04681803658604622, + "kept_token_count": 654.878125, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917245984077454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008275408297777176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8868469238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3915618896484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19250, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20011096745729445, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.12985946461558343, + "grad_norm": 0.050537705421447754, + "kept_token_count": 654.5734375, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.991563904285431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00843609506264329, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0546142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3023376464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19260, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20047229677438735, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.10875311344861985, + "grad_norm": 0.04997272789478302, + "kept_token_count": 654.478125, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916195571422577, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008380455244332553, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.22376403808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3237579345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19270, + "valid_token_count": 818.584375 + }, + { + "actual_token_mask_ratio": 0.1998024344444275, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.1392611362040043, + "grad_norm": 0.0565607026219368, + "kept_token_count": 654.4171875, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912677049636841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008732309658080339, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4075042724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2155242919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19280, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.19988492876291275, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.12866838350892068, + "grad_norm": 0.04450526461005211, + "kept_token_count": 654.60625, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911349236965179, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008865073043853044, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.08070983886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.50006713867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19290, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19964303076267242, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.12785947173833848, + "grad_norm": 0.04745563864707947, + "kept_token_count": 654.846875, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9908065617084503, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009193439316004515, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2292846679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0015075683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19300, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.1990208297967911, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.1264375329017639, + "grad_norm": 0.04818272218108177, + "kept_token_count": 655.5796875, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.991196197271347, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008803788013756276, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4890655517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5442687988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.1998385101556778, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.12691705599427222, + "grad_norm": 0.04603074863553047, + "kept_token_count": 654.6625, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912679851055145, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008732013404369354, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8082214355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.45318908691405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19320, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19956096410751342, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.11727908179163933, + "grad_norm": 0.046044234186410904, + "kept_token_count": 655.05, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925028562545777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007497142720967531, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3726104736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.00730895996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19330, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20060309320688247, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.12435303926467896, + "grad_norm": 0.07059219479560852, + "kept_token_count": 654.1578125, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917346060276031, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008265388943254948, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9977722167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.31724243164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19340, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20049936473369598, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.11430214270949364, + "grad_norm": 0.03464728593826294, + "kept_token_count": 654.1828125, + "learning_rate": 9.204037474105611e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910343825817108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008965625148266553, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.56914672851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.52659912109374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20068605840206147, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.121505768597126, + "grad_norm": 0.06017719954252243, + "kept_token_count": 653.971875, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915896952152252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008410300221294164, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4095184326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.88958435058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19360, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.19877976179122925, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.13066405653953553, + "grad_norm": 0.04753917455673218, + "kept_token_count": 655.5703125, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913750290870667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008624960947781802, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4364501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2079315185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19370, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.20109438449144362, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.13000139370560646, + "grad_norm": 0.03688463196158409, + "kept_token_count": 653.796875, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911689937114716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008831004519015551, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.95929565429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1806274414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19380, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2007657155394554, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.12372750714421273, + "grad_norm": 0.05055231600999832, + "kept_token_count": 654.1609375, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913661003112793, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008633903320878744, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3125854492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.93809509277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.19966185539960862, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.1330141991376877, + "grad_norm": 0.04190143942832947, + "kept_token_count": 655.034375, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915753960609436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008424609899520874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8793426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.02900390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19400, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.20032611936330796, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.11529726088047028, + "grad_norm": 0.03778539225459099, + "kept_token_count": 653.971875, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.992356127500534, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007643872685730457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.52077026367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4573699951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19410, + "valid_token_count": 817.803125 + }, + { + "actual_token_mask_ratio": 0.1998984545469284, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.12425066754221917, + "grad_norm": 0.047883808612823486, + "kept_token_count": 654.4203125, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9912352085113525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008764800336211919, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5119171142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3576965332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19420, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.20008722990751265, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.11707584038376809, + "grad_norm": 0.07434476912021637, + "kept_token_count": 654.3078125, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918079257011414, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008192078303545713, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8111541748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1473388671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19430, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.199602310359478, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.1399785913527012, + "grad_norm": 0.039777446538209915, + "kept_token_count": 654.86875, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9909031391143799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009096871688961982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1145782470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9039825439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19440, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.1991213947534561, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.11625973135232925, + "grad_norm": 0.07026967406272888, + "kept_token_count": 655.38125, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.99164719581604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00835281740874052, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.94281616210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6980865478516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19450, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19989104717969894, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.1280612602829933, + "grad_norm": 0.03888541832566261, + "kept_token_count": 654.415625, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918238461017609, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008176156226545573, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3557556152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.88983154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19460, + "valid_token_count": 817.90625 + }, + { + "actual_token_mask_ratio": 0.19987072199583053, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.12612396776676177, + "grad_norm": 0.09029842913150787, + "kept_token_count": 654.64375, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914564788341522, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008543530013412237, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5563659667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19470, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1996690645813942, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.12872181236743926, + "grad_norm": 0.05645982176065445, + "kept_token_count": 654.74375, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915777683258057, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00842223446816206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5792266845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3763488769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20094446390867232, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.12936163693666458, + "grad_norm": 0.0409376434981823, + "kept_token_count": 654.159375, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915966689586639, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008403323218226432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.85496826171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5555816650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19490, + "valid_token_count": 818.6609375 + }, + { + "actual_token_mask_ratio": 0.19955082535743712, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.12942262589931489, + "grad_norm": 0.042110625654459, + "kept_token_count": 654.9828125, + "learning_rate": 8.976082870365191e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910118877887726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008988119289278984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9036437988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6373352050781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19500, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20085894614458083, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.13059412762522699, + "grad_norm": 0.042834874242544174, + "kept_token_count": 653.834375, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.991685551404953, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008314440492540598, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.16517028808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2702301025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19510, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1998232275247574, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.1373711846768856, + "grad_norm": 0.04970526695251465, + "kept_token_count": 654.54375, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911525905132293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008847418054938317, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2895538330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.64055786132815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19520, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.20015372931957245, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.13322214037179947, + "grad_norm": 0.06752639263868332, + "kept_token_count": 654.484375, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9907452404499054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009254761971533298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6135772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.19281616210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19530, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.20030762404203414, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.13246280401945115, + "grad_norm": 0.07054176926612854, + "kept_token_count": 654.4953125, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9907717168331146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00922828009352088, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.21433715820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0288940429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19540, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2004660487174988, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.1290853001177311, + "grad_norm": 0.0484621524810791, + "kept_token_count": 654.175, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991362202167511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008637805003672838, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.40465087890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.27540588378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19550, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.1997723788022995, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.1239046148955822, + "grad_norm": 0.04964160919189453, + "kept_token_count": 654.959375, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912519156932831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00874808831140399, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8709655761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7713348388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19560, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.19939054399728776, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.1300234951078892, + "grad_norm": 0.043230265378952026, + "kept_token_count": 655.328125, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915420293807984, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008457973692566156, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.11208190917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6612579345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19570, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.20030577033758162, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.13075745105743408, + "grad_norm": 0.06054861471056938, + "kept_token_count": 654.2984375, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912855327129364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008714478090405464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7607666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.56707153320315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19580, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20006328523159028, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.11726566404104233, + "grad_norm": 0.04405755549669266, + "kept_token_count": 654.796875, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918981313705444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008101873844861985, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.88994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.33436279296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19590, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.19997057169675828, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.12186134904623032, + "grad_norm": 0.04593759775161743, + "kept_token_count": 654.7859375, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916730940341949, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008326910156756639, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.31238708496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0194946289063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19600, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.1992507591843605, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.13261644989252092, + "grad_norm": 0.04990454018115997, + "kept_token_count": 655.3203125, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916924118995667, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00830759322270751, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.01946716308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.28285522460936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19610, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20044883489608764, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.12484576925635338, + "grad_norm": 0.06951816380023956, + "kept_token_count": 654.1515625, + "learning_rate": 8.79490196371181e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919779241085053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008022070210427047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0525146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1055633544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19620, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.1997492119669914, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.13793908506631852, + "grad_norm": 0.053783610463142395, + "kept_token_count": 654.8953125, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913478314876556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008652173168957233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5588684082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.00585021972654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19630, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19976236820220947, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.13623154908418655, + "grad_norm": 0.052468735724687576, + "kept_token_count": 654.8015625, + "learning_rate": 8.764809847932277e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910080969333649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00899189105257392, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0884216308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.87586364746096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19640, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20023125261068345, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.1298656977713108, + "grad_norm": 0.046834636479616165, + "kept_token_count": 654.353125, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914983928203582, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008501604385673999, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9710968017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2202667236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19650, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19942569732666016, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.13799562007188798, + "grad_norm": 0.037191081792116165, + "kept_token_count": 655.021875, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911790430545807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008820944838225842, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3619415283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.484033203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19660, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20057913213968276, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.12834578901529312, + "grad_norm": 0.05753908306360245, + "kept_token_count": 654.2703125, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9909257829189301, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009074217081069947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3764923095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.87049865722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19670, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.20056276619434357, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.13365913107991217, + "grad_norm": 0.04611334577202797, + "kept_token_count": 653.953125, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916508555412292, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008349124900996685, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5449737548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.29791870117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19680, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19987233281135558, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.14041176810860634, + "grad_norm": 0.05991087108850479, + "kept_token_count": 654.8734375, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911907970905304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008809191733598709, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6017181396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7401397705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19690, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.1994605079293251, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.12700600773096085, + "grad_norm": 0.04465852677822113, + "kept_token_count": 655.0109375, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917518854141235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008248103596270085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5668151855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14402465820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19700, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2007866993546486, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.12876176312565804, + "grad_norm": 0.04370524734258652, + "kept_token_count": 653.8765625, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917020261287689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00829797675833106, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.98724975585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6292724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19710, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2006254807114601, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.13259364888072014, + "grad_norm": 0.05359210446476936, + "kept_token_count": 654.3265625, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923653662204742, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007634648773819208, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.34666442871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.54306945800784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19720, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.1999162882566452, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.13374357894062996, + "grad_norm": 0.04379194229841232, + "kept_token_count": 654.4390625, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916597902774811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008340206276625396, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7593109130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2395355224609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19730, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.20047955811023713, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.1290617436170578, + "grad_norm": 0.05895562097430229, + "kept_token_count": 653.9515625, + "learning_rate": 8.614806769544507e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919563591480255, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008043649885803462, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3811798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.295654296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19740, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.20044390857219696, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.11680238097906112, + "grad_norm": 0.0467507541179657, + "kept_token_count": 654.2453125, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.991217452287674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008782543055713177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.52903137207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.74614868164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19750, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20119889676570893, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.12261494547128678, + "grad_norm": 0.06019837036728859, + "kept_token_count": 653.5390625, + "learning_rate": 8.584898674669772e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910458087921142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008954181894659996, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.12091064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.99058227539064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19760, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.19870081394910813, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.11877360641956329, + "grad_norm": 0.046494580805301666, + "kept_token_count": 655.6203125, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919444859027863, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00805550692602992, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.36956176757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3481384277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19770, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.20033049285411836, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.12890129238367082, + "grad_norm": 0.04519231989979744, + "kept_token_count": 654.2625, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917829871177674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008217004779726266, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.98992309570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.80482788085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19780, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2009146735072136, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.12793608903884887, + "grad_norm": 0.050760697573423386, + "kept_token_count": 654.0046875, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914707899093628, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008529215771704911, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3469543457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1073791503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19790, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19980574399232864, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.13035386502742768, + "grad_norm": 0.041366949677467346, + "kept_token_count": 654.8296875, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916189730167388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00838102651759982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.12330627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.58306274414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19958445578813552, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.12554244101047515, + "grad_norm": 0.045553918927907944, + "kept_token_count": 655.05, + "learning_rate": 8.510265166731151e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920277655124664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007972247805446387, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.14304504394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3000183105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19810, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2005041792988777, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.13056181594729424, + "grad_norm": 0.05649547278881073, + "kept_token_count": 654.13125, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916191637516022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008380837552249432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5509033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.184326171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19820, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19859346747398376, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.1301462985575199, + "grad_norm": 0.0675404816865921, + "kept_token_count": 655.78125, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914655923843384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008534416649490595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5116027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.88222351074216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19830, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20072001516819, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.12617636397480964, + "grad_norm": 0.08507557958364487, + "kept_token_count": 653.76875, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913814067840576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008618576917797327, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7003204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0410400390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19840, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.20054571479558944, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.1262694254517555, + "grad_norm": 0.057027995586395264, + "kept_token_count": 654.1890625, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912583649158477, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008741641230881215, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.35910949707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.01222534179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19855538308620452, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.12623069509863855, + "grad_norm": 0.051680080592632294, + "kept_token_count": 655.759375, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914476811885834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00855229627341032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.31767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8647521972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20076456069946289, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.1273277811706066, + "grad_norm": 0.04530227929353714, + "kept_token_count": 653.9875, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918096423149109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008190357778221368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5546569824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9270294189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19870, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19848594814538956, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.12437471151351928, + "grad_norm": 0.06421443819999695, + "kept_token_count": 656.040625, + "learning_rate": 8.406109958388222e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910350799560547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008964923862367868, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3289031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.02207336425784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19880, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20013590157032013, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.12639194205403328, + "grad_norm": 0.04198787733912468, + "kept_token_count": 654.3671875, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919213771820068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008078623842447997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.96669616699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.46409606933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19890, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20004622489213944, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.12252749279141426, + "grad_norm": 0.04576241225004196, + "kept_token_count": 654.2921875, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915037453174591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00849625449627638, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5116302490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9130828857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19900, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.20015713423490525, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.1353132866322994, + "grad_norm": 0.04728993773460388, + "kept_token_count": 654.35, + "learning_rate": 8.361591833573612e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9899782359600067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.01002176133915782, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.5321411132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7309875488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20052282363176346, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.1310842141509056, + "grad_norm": 0.05522377789020538, + "kept_token_count": 653.9515625, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913023114204407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008697695098817348, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5655181884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2304229736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19920, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.19997744411230087, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.13140119910240172, + "grad_norm": 0.057962700724601746, + "kept_token_count": 654.6296875, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911162078380584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008883783314377069, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8260192871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9680236816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19930, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.20009004175662995, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.12516319528222083, + "grad_norm": 0.04845086485147476, + "kept_token_count": 654.6234375, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914416790008544, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008558326493948698, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0179779052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5545959472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19940, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.20025432258844375, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.12838054448366165, + "grad_norm": 0.05079787224531174, + "kept_token_count": 654.1921875, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922745823860168, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007725418917834759, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5181915283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.98480834960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19950, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.19962004572153091, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.1321167081594467, + "grad_norm": 0.06750930845737457, + "kept_token_count": 654.9125, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991414487361908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008585509844124317, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.60703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.31529541015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19960, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19962807446718217, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.11998372226953506, + "grad_norm": 0.06716371327638626, + "kept_token_count": 654.8890625, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921609699726105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007839027792215347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9277404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9571533203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19970, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19981749057769777, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.1288231447339058, + "grad_norm": 0.0495651513338089, + "kept_token_count": 654.5734375, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9908377230167389, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009162271209061146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.11668395996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.1457458496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19980, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.19998528212308883, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.12014267444610596, + "grad_norm": 0.04088297858834267, + "kept_token_count": 654.5609375, + "learning_rate": 8.243233070395266e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910204648971558, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008979546651244164, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4813232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7765350341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19990, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.1998603269457817, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.1270347572863102, + "grad_norm": 0.04512433707714081, + "kept_token_count": 654.5375, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917704999446869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008229510113596916, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8180419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4451416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20000, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.19957332611083983, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.115735724568367, + "grad_norm": 0.06160468980669975, + "kept_token_count": 654.9265625, + "learning_rate": 8.213725121518346e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910484671592712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008951522875577211, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6783416748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7840850830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20010, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2004026100039482, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.1193207859992981, + "grad_norm": 0.06112327054142952, + "kept_token_count": 654.184375, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915916860103607, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408299926668406, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0970825195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1991790771484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20020, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2006424769759178, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.12630897164344787, + "grad_norm": 0.04479407146573067, + "kept_token_count": 654.0125, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916246354579925, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00837536882609129, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2687469482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15419616699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20030, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.20007889717817307, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.1176015391945839, + "grad_norm": 0.07630383223295212, + "kept_token_count": 654.5203125, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916411638259888, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008358843438327313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3131378173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.61211853027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20040, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2000088468194008, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.1160295031964779, + "grad_norm": 0.0390591025352478, + "kept_token_count": 654.6953125, + "learning_rate": 8.154808318560413e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919768512248993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008023140858858824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7824462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.62481079101565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20050, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.19983571916818618, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.12432299479842186, + "grad_norm": 0.04686694219708443, + "kept_token_count": 654.690625, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.991478317975998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008521680999547244, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.60279235839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.783447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20060, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.20012587457895278, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.11603052765130997, + "grad_norm": 0.04015707969665527, + "kept_token_count": 654.409375, + "learning_rate": 8.125399750837202e-06, + "loss": 0.008, + "mgd_cos_sim": 0.992033737897873, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00796626452356577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.21726379394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1949920654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20070, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20002121925354005, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.12381356209516525, + "grad_norm": 0.07384880632162094, + "kept_token_count": 654.4578125, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912979543209076, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0087020474486053, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.10407104492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09075927734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20080, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2001548543572426, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.1215054601430893, + "grad_norm": 0.05434263497591019, + "kept_token_count": 654.6015625, + "learning_rate": 8.096024596253982e-06, + "loss": 0.008, + "mgd_cos_sim": 0.992028659582138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007971340045332908, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.52532958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.34454040527345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20090, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.20001388490200042, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.12028948217630386, + "grad_norm": 0.04773702844977379, + "kept_token_count": 654.6796875, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917495965957641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008250411693006754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.42437438964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2877166748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20100, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19925581961870192, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.11857519671320915, + "grad_norm": 0.04469845071434975, + "kept_token_count": 655.2265625, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9908903002738952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009109696466475725, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.75812377929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6151947021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19974194914102555, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.12348071187734604, + "grad_norm": 0.050214119255542755, + "kept_token_count": 654.903125, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991417545080185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008582468517124653, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.01953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7728515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19988596588373184, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.12491438388824463, + "grad_norm": 0.04826286807656288, + "kept_token_count": 654.5703125, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916482508182526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008351753652095794, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8540313720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9654510498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20130, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19940169155597687, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.12798659652471542, + "grad_norm": 0.08666316419839859, + "kept_token_count": 655.0734375, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916692614555359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008330743201076985, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.27586975097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.21392211914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20140, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2003932312130928, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.11816587075591087, + "grad_norm": 0.049018219113349915, + "kept_token_count": 654.146875, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911868274211884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00881317276507616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4292419433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.94228210449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20150, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.19913479536771775, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.1222475066781044, + "grad_norm": 0.039832741022109985, + "kept_token_count": 655.1828125, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.991463965177536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008536040876060724, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.79042663574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.23938903808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20160, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20011359602212905, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.12161222174763679, + "grad_norm": 0.05945035070180893, + "kept_token_count": 654.8046875, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916560471057891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008343951031565666, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7870300292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2866668701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20170, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.1998979702591896, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.11476442739367484, + "grad_norm": 0.04249318689107895, + "kept_token_count": 654.6359375, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914861023426056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008513911813497543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3411865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8832611083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20180, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.1993790552020073, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.12150350734591484, + "grad_norm": 0.06268688291311264, + "kept_token_count": 655.2125, + "learning_rate": 7.949655011814968e-06, + "loss": 0.009, + "mgd_cos_sim": 0.991034185886383, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0089658098295331, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9421417236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1232635498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20190, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.1992407262325287, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.12187740355730056, + "grad_norm": 0.055661190301179886, + "kept_token_count": 655.01875, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918859601020813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008114043623209, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6926025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.74817810058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20200, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.1989261507987976, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.13543706834316255, + "grad_norm": 0.047718074172735214, + "kept_token_count": 655.4203125, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.9907928347587586, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009207166824489832, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5767395019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4677490234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20210, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19983375817537308, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.13143920972943307, + "grad_norm": 0.06409268826246262, + "kept_token_count": 654.8671875, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911220908164978, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008877908810973167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3102630615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0599609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20220, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.20058366358280183, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.12727445885539054, + "grad_norm": 0.07397899776697159, + "kept_token_count": 654.1015625, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913311779499054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008668829035013915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.63486633300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.47556762695314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20230, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.19978825747966766, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.11811173185706139, + "grad_norm": 0.05653941258788109, + "kept_token_count": 654.7546875, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917311429977417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008268861938267947, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8164794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25185241699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20240, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19948810040950776, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.12270583510398865, + "grad_norm": 0.06786705553531647, + "kept_token_count": 654.7796875, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916768372058868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008323155716061592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.418505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.09313049316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20250, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.2018658325076103, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.1398064211010933, + "grad_norm": 0.04638436436653137, + "kept_token_count": 653.021875, + "learning_rate": 7.847704970163127e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920206606388092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007979335822165012, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.17372436523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.10532531738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20260, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20046715885400773, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.11559421494603157, + "grad_norm": 0.09073309600353241, + "kept_token_count": 654.271875, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915274024009705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008472609799355268, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6664703369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8025848388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20270, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.20047179758548736, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.11635238453745841, + "grad_norm": 0.05148869752883911, + "kept_token_count": 654.2046875, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915586411952972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008441348653286695, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0150665283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9153594970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20280, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20014441311359404, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.11838757917284966, + "grad_norm": 0.039826683700084686, + "kept_token_count": 654.4796875, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918687343597412, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00813126852735877, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8928741455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.76085205078124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20290, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19982908070087432, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.13208127096295358, + "grad_norm": 0.03745220601558685, + "kept_token_count": 654.771875, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991382485628128, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00861751800402999, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.68976135253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1617462158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20300, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20033951848745346, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.13256967812776566, + "grad_norm": 0.05286995321512222, + "kept_token_count": 654.3875, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9912061035633087, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008793883491307497, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2027862548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.80699768066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2006834328174591, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.12348895668983459, + "grad_norm": 0.04816574230790138, + "kept_token_count": 654.1703125, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918210208415985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008178982511162758, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.05613403320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40809020996096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2011708974838257, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.12368318662047387, + "grad_norm": 0.041374772787094116, + "kept_token_count": 653.8515625, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920931458473206, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007906859554350377, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8558654785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.67458190917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20330, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.19970585107803346, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.12426964789628983, + "grad_norm": 0.043677084147930145, + "kept_token_count": 655.0421875, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918739676475525, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008126023691147566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.90374145507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4905944824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20340, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2006659150123596, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.12902424931526185, + "grad_norm": 0.04053013026714325, + "kept_token_count": 654.0328125, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9910865008831025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008913513738662004, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8175048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.886474609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.200672011077404, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.12061119228601455, + "grad_norm": 0.046848099678754807, + "kept_token_count": 654.034375, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922901332378388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007709862384945154, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.26634826660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9460968017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20360, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19978947937488556, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.1284516856074333, + "grad_norm": 0.045066751539707184, + "kept_token_count": 654.44375, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921238422393799, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007876156177371741, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.62901306152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.06480712890624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20370, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.1995534896850586, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.13784475177526473, + "grad_norm": 0.04131993278861046, + "kept_token_count": 654.99375, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913593292236328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00864067319780588, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.07399291992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0585662841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20380, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19910721778869628, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.11729835644364357, + "grad_norm": 0.04280438274145126, + "kept_token_count": 655.2859375, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918871939182281, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008112794812768697, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.89251098632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.91318664550784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20390, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.20010701715946197, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.1285279542207718, + "grad_norm": 0.04399125277996063, + "kept_token_count": 654.4203125, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918460488319397, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008153935428708792, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3532745361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.37913513183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20400, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.20053440779447557, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.12581675723195077, + "grad_norm": 0.05749369412660599, + "kept_token_count": 654.1359375, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911966800689698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008803332690149546, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.29523315429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.11911926269534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20410, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.20044868290424347, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.12879957556724547, + "grad_norm": 0.06259024143218994, + "kept_token_count": 654.3359375, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919953525066376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008004628587514162, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.94679565429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.00458068847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20420, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19952699095010756, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.1269419878721237, + "grad_norm": 0.041517745703458786, + "kept_token_count": 654.903125, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923927545547485, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00760723352432251, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.975244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9576843261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20430, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.19993216544389725, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.13095874190330506, + "grad_norm": 0.05064539611339569, + "kept_token_count": 654.5578125, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916434168815613, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008356573246419429, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1364379882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41156311035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20440, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20046332180500032, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.11772817149758338, + "grad_norm": 0.042294442653656006, + "kept_token_count": 654.3234375, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.991839337348938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008160665910691024, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0434814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2698455810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20450, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19919052720069885, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.12233124300837517, + "grad_norm": 0.0483505055308342, + "kept_token_count": 655.071875, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913840472698212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008615954872220754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.40151062011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5306121826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20460, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.20008278638124466, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.12745083793997763, + "grad_norm": 0.048108041286468506, + "kept_token_count": 654.29375, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913021147251129, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008697882760316133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.15386962890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6945007324219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20470, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.19972880631685258, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.14243075102567673, + "grad_norm": 0.03685645014047623, + "kept_token_count": 654.8140625, + "learning_rate": 7.530090951955334e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920139908790588, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007986019551753997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.23761596679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.96592407226564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20480, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20020930618047714, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.12895672097802163, + "grad_norm": 0.04413646087050438, + "kept_token_count": 654.56875, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913832128047944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008616781793534756, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.47313232421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.82704162597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20490, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.1988375723361969, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.11897326931357384, + "grad_norm": 0.053468991070985794, + "kept_token_count": 655.434375, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915431857109069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008456816058605909, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.17564086914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.21385498046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20500, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.20066384226083755, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.12520990073680877, + "grad_norm": 0.05964547395706177, + "kept_token_count": 653.8234375, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912602186203003, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008739768154919147, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4198974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.00023193359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20510, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.20094472914934158, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.11681079491972923, + "grad_norm": 0.04936248064041138, + "kept_token_count": 653.825, + "learning_rate": 7.472809478761672e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919622302055359, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008037766814231873, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.975, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.19794311523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1992679938673973, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.12371457219123841, + "grad_norm": 0.048535726964473724, + "kept_token_count": 655.1234375, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914576768875122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008542312309145927, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1880340576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.96690368652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20530, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1997288480401039, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.129538207501173, + "grad_norm": 0.04847189784049988, + "kept_token_count": 654.8078125, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0088, + "mgd_cos_sim": 0.9911938011646271, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00880620051175356, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3830932617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19556579589846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20540, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.1997840851545334, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.11761357858777047, + "grad_norm": 0.04888523742556572, + "kept_token_count": 655.140625, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.991918396949768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008081607054919004, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3129119873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.57520446777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20550, + "valid_token_count": 818.70625 + }, + { + "actual_token_mask_ratio": 0.19977979212999344, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.12527824565768242, + "grad_norm": 0.04653702676296234, + "kept_token_count": 654.759375, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.9908812284469605, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009118771366775035, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9247802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30919189453124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20560, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19958298057317733, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.1174190379679203, + "grad_norm": 0.045913320034742355, + "kept_token_count": 655.0671875, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913417279720307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008658254705369472, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.5842681884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9281768798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20570, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.20038577169179916, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.12610457092523575, + "grad_norm": 0.06983257085084915, + "kept_token_count": 654.1828125, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913831651210785, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008616832736879586, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9291595458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.22289123535154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20580, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20041301101446152, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.12217446863651275, + "grad_norm": 0.04078267887234688, + "kept_token_count": 654.0, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924963653087616, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007503636740148067, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.907470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.98798828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20590, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.20003016591072081, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.13086874783039093, + "grad_norm": 0.042449213564395905, + "kept_token_count": 654.4765625, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920873939990997, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007912611123174429, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.4523986816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3856903076172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20600, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.20052071809768676, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.12761091366410254, + "grad_norm": 0.036679573357105255, + "kept_token_count": 654.284375, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.99148810505867, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008511894475668668, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1787811279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.45628356933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20610, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20008163154125214, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.12531451880931854, + "grad_norm": 0.06417187303304672, + "kept_token_count": 654.659375, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.991635799407959, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008364210370928049, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9203857421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4960876464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20620, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20050528794527053, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.1250136002898216, + "grad_norm": 0.05804768577218056, + "kept_token_count": 654.1546875, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914708375930786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008529163990169763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9878662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20630, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20055443048477173, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.13012745678424836, + "grad_norm": 0.04932495579123497, + "kept_token_count": 654.0265625, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917667210102081, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008233290910720826, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4425506591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.08055419921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.20019040256738663, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.1135724775493145, + "grad_norm": 0.04238288849592209, + "kept_token_count": 654.353125, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915505290031433, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008449482079595328, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.69858703613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0427612304687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20650, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19944499731063842, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.13001967072486878, + "grad_norm": 0.05914103612303734, + "kept_token_count": 655.353125, + "learning_rate": 7.273484030370423e-06, + "loss": 0.009, + "mgd_cos_sim": 0.9910455346107483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00895446864888072, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9967834472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.17368469238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20660, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20011309683322906, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.12020484805107116, + "grad_norm": 0.05545327067375183, + "kept_token_count": 654.79375, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917591869831085, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008240816835314035, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2537780761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3401336669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20670, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.2004668429493904, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.1314733639359474, + "grad_norm": 0.0472392775118351, + "kept_token_count": 653.9546875, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913023293018342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008697674237191677, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.44301147460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6601989746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20680, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.20006545037031173, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.12380203530192375, + "grad_norm": 0.05129045993089676, + "kept_token_count": 654.80625, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916030108928681, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008396982494741679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7984893798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05401916503905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20690, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.20101188868284225, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.12293460667133331, + "grad_norm": 0.043091509491205215, + "kept_token_count": 653.8296875, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0093, + "mgd_cos_sim": 0.9906984031200409, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009301605075597763, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.29823913574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9506103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.19973042011260986, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.1185978539288044, + "grad_norm": 0.05871589481830597, + "kept_token_count": 654.884375, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915994942188263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008400505222380162, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.88013610839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8577514648438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20710, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2003808155655861, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.12645957097411156, + "grad_norm": 0.04731869697570801, + "kept_token_count": 654.365625, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917145609855652, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008285415172576905, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.78186950683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9502746582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20720, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.199961094558239, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.1261713370680809, + "grad_norm": 0.055242713540792465, + "kept_token_count": 654.8078125, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924298703670502, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007570127677172423, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.31646118164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.17127990722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20730, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.19932232350111007, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.12121163234114647, + "grad_norm": 0.04458519071340561, + "kept_token_count": 655.171875, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925560116767883, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00744399856775999, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.87749633789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1960754394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20740, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20053675919771194, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.12347248867154122, + "grad_norm": 0.04264200106263161, + "kept_token_count": 654.365625, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992176228761673, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007823776826262474, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7903259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9938201904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20750, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19977195262908937, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.11903353929519653, + "grad_norm": 0.06297765672206879, + "kept_token_count": 654.7328125, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9910779178142548, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008922073617577552, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.60854187011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9888977050781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20760, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19995518028736115, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.1238170936703682, + "grad_norm": 0.07486779987812042, + "kept_token_count": 654.715625, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918549060821533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008145103044807912, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.86346130371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8339813232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20770, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19987425655126573, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.12199191749095917, + "grad_norm": 0.07135780900716782, + "kept_token_count": 655.0390625, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916293323040009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008370657730847598, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7091430664062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.06842346191405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20780, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.19951325058937072, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.12393120974302292, + "grad_norm": 0.03246680647134781, + "kept_token_count": 654.834375, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921638369560242, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00783615866675973, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7839782714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6962097167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20790, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.19958497285842897, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.12066265791654587, + "grad_norm": 0.08893778920173645, + "kept_token_count": 654.8, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920507192611694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007949272729456424, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2122741699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.268310546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20800, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.1994430363178253, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.1295418567955494, + "grad_norm": 0.0484045185148716, + "kept_token_count": 655.221875, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913753211498261, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00862467773258686, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3648315429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0260803222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20810, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.20013059675693512, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.12029601410031318, + "grad_norm": 0.04439358785748482, + "kept_token_count": 654.3265625, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916555941104889, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00834441240876913, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6679229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83629150390624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20820, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2001433178782463, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.12663176655769348, + "grad_norm": 0.06084488332271576, + "kept_token_count": 654.63125, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912937462329865, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008706254325807095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1038391113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8129852294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19994807988405228, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.1275398813188076, + "grad_norm": 0.04314703866839409, + "kept_token_count": 654.646875, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919174015522003, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008082589879631997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.32506408691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.42412719726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20840, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20044273138046265, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.11105310395359994, + "grad_norm": 0.0498325452208519, + "kept_token_count": 654.2125, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918715000152588, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00812849411740899, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4426239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.78830871582034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20850, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19965071380138397, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.13006228357553482, + "grad_norm": 0.06007232517004013, + "kept_token_count": 654.753125, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924420118331909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00755798676982522, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9814025878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3594024658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20860, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2002093330025673, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.12270739823579788, + "grad_norm": 0.040292493999004364, + "kept_token_count": 654.4609375, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917929351329804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008207066077739, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.00017700195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4379150390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20870, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2002584904432297, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.11760636419057846, + "grad_norm": 0.04025823250412941, + "kept_token_count": 654.31875, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924242079257966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007575792632997036, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.06019287109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7305480957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20880, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20031503587961197, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.13159079030156134, + "grad_norm": 0.06943590193986893, + "kept_token_count": 654.4625, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915280997753143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008471909910440445, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8245147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0416229248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20890, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19922335147857667, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.13245279863476753, + "grad_norm": 0.042376790195703506, + "kept_token_count": 655.4890625, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916232764720917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008376731723546981, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4879913330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4727752685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20900, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.1994609132409096, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.12076012045145035, + "grad_norm": 0.05679160729050636, + "kept_token_count": 655.0703125, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.991503757238388, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008496229536831379, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5894073486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.55311279296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20910, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19952923506498338, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.13700857758522034, + "grad_norm": 0.04468371346592903, + "kept_token_count": 655.0453125, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921736240386962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00782637931406498, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.47789001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6004943847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20920, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19914479106664656, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.11461569517850875, + "grad_norm": 0.045473210513591766, + "kept_token_count": 655.61875, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917900443077088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00820995308458805, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7349853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8529846191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20930, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.20022473186254502, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.12406045123934746, + "grad_norm": 0.04141572490334511, + "kept_token_count": 654.425, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916237235069275, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00837628087028861, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9370361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3571075439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20940, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19921600967645645, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.1288813643157482, + "grad_norm": 0.05059172213077545, + "kept_token_count": 655.1140625, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919433653354645, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008056631870567798, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9985595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.28936767578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20063867419958115, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.13431726694107055, + "grad_norm": 0.045777130872011185, + "kept_token_count": 654.0640625, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918919682502747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008108009677380323, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.47817077636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.60537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20960, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19960611462593078, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.1283629171550274, + "grad_norm": 0.051788512617349625, + "kept_token_count": 654.99375, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914857983589173, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008514204528182745, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.11915893554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.30304565429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20970, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19897091388702393, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.12603959813714027, + "grad_norm": 0.059037819504737854, + "kept_token_count": 655.6546875, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916602432727813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00833974927663803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4857879638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.25587158203126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20980, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.20005950331687927, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.12040252462029458, + "grad_norm": 0.048617493361234665, + "kept_token_count": 654.409375, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921160638332367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00788393896073103, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.25177307128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48646240234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20990, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20051027834415436, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.12041479572653771, + "grad_norm": 0.0417233407497406, + "kept_token_count": 654.13125, + "learning_rate": 6.797161907509499e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920437514781952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00795625764876604, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.656689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3536437988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21000, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.20014435350894927, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.13267949149012565, + "grad_norm": 0.05049654468894005, + "kept_token_count": 654.5265625, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921059250831604, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00789409140124917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5321502685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4278259277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.20084301978349686, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.12986998707056047, + "grad_norm": 0.05944550037384033, + "kept_token_count": 653.66875, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911407947540283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008859221078455447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.89723205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94756774902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21020, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.2004931852221489, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.12352040782570839, + "grad_norm": 0.056390389800071716, + "kept_token_count": 654.4015625, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916160106658936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008383985422551633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.8244689941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0315307617187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21030, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.19963106513023376, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.1267300069332123, + "grad_norm": 0.061304330825805664, + "kept_token_count": 654.884375, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917740046977996, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008225984871387482, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0702392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7991424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21040, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19970519542694093, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.12266543507575989, + "grad_norm": 0.06563049554824829, + "kept_token_count": 654.8765625, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914814174175263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008518578950315714, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0579376220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9870269775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21050, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.19993523955345155, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.12072965130209923, + "grad_norm": 0.05366190895438194, + "kept_token_count": 654.65625, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921333074569703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007866691052913665, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7914215087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3354187011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21060, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19943425357341765, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.11600907668471336, + "grad_norm": 0.05467041954398155, + "kept_token_count": 655.1046875, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920949876308441, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007905003055930138, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.00331115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6846008300781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21070, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.20062065720558167, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.1283445529639721, + "grad_norm": 0.052862975746393204, + "kept_token_count": 654.2109375, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0091, + "mgd_cos_sim": 0.990864223241806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009135782066732646, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.64105834960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5583221435547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21080, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19931880235671998, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.1214709885418415, + "grad_norm": 0.05064063146710396, + "kept_token_count": 655.1640625, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920972168445588, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007902780920267105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.76845092773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0793701171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21090, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19960662126541137, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.12439251616597176, + "grad_norm": 0.05377820506691933, + "kept_token_count": 654.9890625, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919359028339386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008064108900725842, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2685302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32838745117186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21100, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19910566955804826, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.12710001245141028, + "grad_norm": 0.04631320387125015, + "kept_token_count": 655.225, + "learning_rate": 6.645492438179188e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919940054416656, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008005972765386105, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.61570129394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6071258544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21110, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.20106492638587953, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.11931359991431237, + "grad_norm": 0.05817679688334465, + "kept_token_count": 653.875, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913099408149719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008690056297928095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.30894775390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.44585876464845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21120, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.19978984147310258, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.13029008060693742, + "grad_norm": 0.055914659053087234, + "kept_token_count": 654.9109375, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915810763835907, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008418924733996391, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.86590881347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3779235839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21130, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.2003002032637596, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.1204369343817234, + "grad_norm": 0.040310535579919815, + "kept_token_count": 654.1375, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.991634601354599, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008365391474217176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.11365661621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5574188232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21140, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.199869567155838, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.12685653790831566, + "grad_norm": 0.049316681921482086, + "kept_token_count": 654.565625, + "learning_rate": 6.590643278432342e-06, + "loss": 0.008, + "mgd_cos_sim": 0.992016053199768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00798395685851574, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4364410400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8753997802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21150, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.20004269182682038, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.11136310026049615, + "grad_norm": 0.036274366080760956, + "kept_token_count": 654.428125, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916774451732635, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008322562277317046, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6007934570313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40862426757815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19944607317447663, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.12008554637432098, + "grad_norm": 0.04570712149143219, + "kept_token_count": 655.3546875, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917981505393982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008201852347701788, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.36663818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.35431213378905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21170, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.20011761486530305, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.12606502175331116, + "grad_norm": 0.09491878002882004, + "kept_token_count": 654.6390625, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916127860546112, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008387218788266182, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5665802001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.72106018066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19970825165510178, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.12161132246255875, + "grad_norm": 0.06378185749053955, + "kept_token_count": 654.7078125, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915913701057434, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408630546182395, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.45665283203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.00208435058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21190, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20050466060638428, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.11647749915719033, + "grad_norm": 0.040273409336805344, + "kept_token_count": 654.36875, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916638135910034, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00833618426695466, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4499481201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3461212158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21200, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20186958760023116, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.1282760888338089, + "grad_norm": 0.06030561029911041, + "kept_token_count": 652.9515625, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911375761032104, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008862415514886379, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9067749023437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1253662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21210, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20039250701665878, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.13023098483681678, + "grad_norm": 0.042401280254125595, + "kept_token_count": 654.290625, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914643704891205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00853562168776989, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.72083435058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5330871582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19986517131328582, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.12355601266026497, + "grad_norm": 0.06558182090520859, + "kept_token_count": 654.49375, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915772438049316, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008422768954187632, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.71180725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.17279663085935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21230, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2006428763270378, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.12809116169810295, + "grad_norm": 0.05294610932469368, + "kept_token_count": 654.2015625, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921958863735199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007804120052605867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.71921691894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1764404296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21240, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20014473795890808, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.12025772109627723, + "grad_norm": 0.052015818655490875, + "kept_token_count": 654.4875, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916462421417236, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008353756368160247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1039154052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.21393432617185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21250, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19954510182142257, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.12256060466170311, + "grad_norm": 0.05435854196548462, + "kept_token_count": 654.9171875, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914640367031098, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008535966277122498, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2518005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.847802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21260, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.19948890656232834, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.12422913610935211, + "grad_norm": 0.04868554323911667, + "kept_token_count": 654.834375, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918087422847748, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008191269356757402, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4098205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8098999023438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21270, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20059817433357238, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.12435130402445793, + "grad_norm": 0.03633231669664383, + "kept_token_count": 654.03125, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924412608146668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007558736018836498, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.32103576660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9726989746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21280, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20029788911342622, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.1286521941423416, + "grad_norm": 0.04397360235452652, + "kept_token_count": 654.5015625, + "learning_rate": 6.399965674872228e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919975101947784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008002477139234543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.82135314941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.95990600585935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21290, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.1992074131965637, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.12641093581914903, + "grad_norm": 0.043083302676677704, + "kept_token_count": 655.3046875, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917408943176269, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008259095251560211, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.81629943847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5743469238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21300, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19911422431468964, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.1290063090622425, + "grad_norm": 0.0569610521197319, + "kept_token_count": 655.365625, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919361770153046, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00806382717564702, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.73316650390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.08543090820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21310, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.2004045680165291, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.11748773604631424, + "grad_norm": 0.05455818399786949, + "kept_token_count": 654.3015625, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917434751987457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0082565245218575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6141296386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14173889160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21320, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19854716807603837, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.12361238822340966, + "grad_norm": 0.051110006868839264, + "kept_token_count": 655.88125, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920927584171295, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0079072343185544, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3111236572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.311572265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21330, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2001076340675354, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.13711495101451873, + "grad_norm": 0.04605473205447197, + "kept_token_count": 654.2703125, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917453646659851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008254628162831068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3419464111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0600128173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21340, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.19994510859251022, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.1376925751566887, + "grad_norm": 0.0942225307226181, + "kept_token_count": 654.6390625, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911101877689361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008889816608279944, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8197448730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2843872070313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.19918326288461685, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.13378777205944062, + "grad_norm": 0.04656782001256943, + "kept_token_count": 655.14375, + "learning_rate": 6.305392102860924e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920400738716125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007959930691868067, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.96918334960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.89708557128904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21360, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.1997773066163063, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.11808729022741318, + "grad_norm": 0.06951986998319626, + "kept_token_count": 654.715625, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911186814308166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008881329372525216, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1649871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0317901611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21370, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19890570044517517, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.13148973807692527, + "grad_norm": 0.06710062175989151, + "kept_token_count": 655.546875, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914638102054596, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008536183647811413, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5251434326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6244323730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21380, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19969528913497925, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.1259221076965332, + "grad_norm": 0.04397548362612724, + "kept_token_count": 654.74375, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913784086704254, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008621589932590722, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6581756591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1951141357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.1996062532067299, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.13257865309715272, + "grad_norm": 0.041996169835329056, + "kept_token_count": 654.871875, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0096, + "mgd_cos_sim": 0.9904172360897064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.009582758229225875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.62437744140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.39085693359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21400, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.2003632128238678, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.12262609302997589, + "grad_norm": 0.05639152601361275, + "kept_token_count": 654.1515625, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914695501327515, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008530453033745288, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.62880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.31430969238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.1997920021414757, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.12669986560940744, + "grad_norm": 0.044252462685108185, + "kept_token_count": 654.5953125, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917237401008606, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008276244159787893, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.52227478027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.58037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21420, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.20037667751312255, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.12034653425216675, + "grad_norm": 0.04944261536002159, + "kept_token_count": 654.35625, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.991734254360199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008265743404626847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.6541748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5612518310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21430, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20092063546180725, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.11987217664718627, + "grad_norm": 0.061720915138721466, + "kept_token_count": 653.815625, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918367862701416, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0081632143817842, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.41681518554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.27069091796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21440, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20029376596212387, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.12270146608352661, + "grad_norm": 0.044125910848379135, + "kept_token_count": 654.5, + "learning_rate": 6.184558954358475e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920426428318023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00795735539868474, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7049957275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9879180908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21450, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.20089295506477356, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.1194703958928585, + "grad_norm": 0.049374282360076904, + "kept_token_count": 653.8875, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922625064849854, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007737494260072708, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.71392211914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5254760742188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21460, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20016349405050277, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.119270658493042, + "grad_norm": 0.051549557596445084, + "kept_token_count": 654.5359375, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918063879013062, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008193617966026068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7259063720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3192169189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21470, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.2003648579120636, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.12732600793242455, + "grad_norm": 0.048223335295915604, + "kept_token_count": 654.275, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914402008056641, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008559793699532747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.25160217285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8408660888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21480, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.19963183552026748, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.12952519729733467, + "grad_norm": 0.0460832454264164, + "kept_token_count": 654.6984375, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921844840049744, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00781551692634821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5262939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63681945800784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21490, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2002563402056694, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.12119989693164826, + "grad_norm": 0.03852204605937004, + "kept_token_count": 654.3390625, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921239495277405, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007876051776111125, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6391845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.58605041503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21500, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2003217190504074, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.12769706025719643, + "grad_norm": 0.06414327770471573, + "kept_token_count": 654.3953125, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992195439338684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00780457304790616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.55792846679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.63446044921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21510, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19891712814569473, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.12567155212163925, + "grad_norm": 0.06789928674697876, + "kept_token_count": 655.571875, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916146576404572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008385339006781577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.532275390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53033447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21520, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19921261072158813, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.12229079753160477, + "grad_norm": 0.06507928669452667, + "kept_token_count": 655.503125, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922603130340576, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077397007495164875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.925537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98786010742185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21530, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.19942864179611205, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.12291873395442962, + "grad_norm": 0.056801822036504745, + "kept_token_count": 655.1625, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913009703159332, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008699024375528098, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6056884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6902709960938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21540, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20035802572965622, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.1308935098350048, + "grad_norm": 0.045221615582704544, + "kept_token_count": 654.2328125, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915475606918335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008452449180185794, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1694274902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5800323486328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21550, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1996331363916397, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.125627451390028, + "grad_norm": 0.0494922399520874, + "kept_token_count": 655.0734375, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922275006771087, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007772504352033138, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0815002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.80491027832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21560, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.1996508240699768, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.12367020845413208, + "grad_norm": 0.04354911670088768, + "kept_token_count": 654.9671875, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.991513055562973, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00848693335428834, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.244482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3681671142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21570, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.20020217448472977, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.12474833503365516, + "grad_norm": 0.03695473447442055, + "kept_token_count": 654.4953125, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924582481384278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007541750371456146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5418243408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9073211669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21580, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19994729310274123, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.1226797454059124, + "grad_norm": 0.051633190363645554, + "kept_token_count": 654.7375, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922860324382782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007713966816663742, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6911346435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7410614013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21590, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20046924203634262, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.12616647630929947, + "grad_norm": 0.043167926371097565, + "kept_token_count": 654.325, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921144723892212, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007885527797043324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7426055908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9870941162109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21600, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20008374005556107, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.12928588837385177, + "grad_norm": 0.05010312423110008, + "kept_token_count": 654.2625, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991437965631485, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008562043029814958, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8787384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.72145690917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21610, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.20073304325342178, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.12036400213837624, + "grad_norm": 0.03748075291514397, + "kept_token_count": 654.109375, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921724677085877, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007827519904822111, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.369287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22926330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21620, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20096722096204758, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.11306687742471695, + "grad_norm": 0.06686443090438843, + "kept_token_count": 653.9921875, + "learning_rate": 5.945507360378608e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920357227325439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007964271120727062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9230560302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0659484863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21630, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20128608047962188, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.11727239936590195, + "grad_norm": 0.054397448897361755, + "kept_token_count": 653.3953125, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992150354385376, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007849638722836972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.116748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0305603027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21640, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.19946223646402358, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.1250789314508438, + "grad_norm": 0.048729609698057175, + "kept_token_count": 655.3109375, + "learning_rate": 5.919164398371857e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920050919055938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007994919270277023, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3795989990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5723114013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21650, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19998181462287903, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.1257498748600483, + "grad_norm": 0.048031050711870193, + "kept_token_count": 654.6671875, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923761188983917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007623888179659843, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3543670654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.24201049804685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21660, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.19967930763959885, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.1307620622217655, + "grad_norm": 0.04597454145550728, + "kept_token_count": 654.525, + "learning_rate": 5.892865572637789e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920199751853943, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007980044279247523, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5789398193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4324249267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21670, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.20083924680948256, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.13437335714697837, + "grad_norm": 0.046421315521001816, + "kept_token_count": 654.0265625, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914526104927063, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008547391276806593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.65692138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.41209411621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21680, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.20019986629486083, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.11281634345650673, + "grad_norm": 0.04368392005562782, + "kept_token_count": 654.6, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921049833297729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007895016577094793, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7082855224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.43174438476564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21690, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.20009519904851913, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.12926043048501015, + "grad_norm": 0.04712527617812157, + "kept_token_count": 654.64375, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916326582431794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008367339335381984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.011865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.24888305664064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21700, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19939043670892714, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.1266883336007595, + "grad_norm": 0.04434695467352867, + "kept_token_count": 655.075, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913210690021514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008678935933858156, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8780181884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9337158203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19992753714323044, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.12261788696050643, + "grad_norm": 0.06399346888065338, + "kept_token_count": 654.725, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924107193946838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007589279301464557, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8931396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1610992431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21720, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.20009354650974273, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.1300822414457798, + "grad_norm": 0.07700424641370773, + "kept_token_count": 654.2578125, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917238235473633, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008276167977601289, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.54767150878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.79163818359376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21730, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.19979617446660997, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.1219877578318119, + "grad_norm": 0.040729351341724396, + "kept_token_count": 654.7640625, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923119068145752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007688092160969972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.43177490234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.12477722167966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21740, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20042202472686768, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.12494814172387123, + "grad_norm": 0.06355088949203491, + "kept_token_count": 654.059375, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922278463840485, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007772145792841911, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6553497314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1623260498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21750, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2001403585076332, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.12195534631609917, + "grad_norm": 0.04431943967938423, + "kept_token_count": 654.60625, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918654561042786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008134547341614961, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.30479125976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.67125244140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21760, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20097564309835433, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.12468351945281028, + "grad_norm": 0.058100976049900055, + "kept_token_count": 653.8140625, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921535611152649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007846433017402888, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.21917114257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8725891113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21770, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19950155466794967, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.1246705949306488, + "grad_norm": 0.06705678999423981, + "kept_token_count": 654.9921875, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917661666870117, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008233831264078617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9243927001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9169586181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21780, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20062079429626464, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.1161257840692997, + "grad_norm": 0.04797597974538803, + "kept_token_count": 654.1484375, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923110365867615, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007688960991799831, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5092529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.89398498535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19956768602132796, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.12437524423003196, + "grad_norm": 0.05712539702653885, + "kept_token_count": 654.8765625, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914341688156127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008565820474177598, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.29332275390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.672998046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21800, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.19903873056173324, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.12177048549056053, + "grad_norm": 0.03366510942578316, + "kept_token_count": 655.4640625, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920500457286835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00794995715841651, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.85007934570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.03785705566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2012360066175461, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.12602950930595397, + "grad_norm": 0.05795472487807274, + "kept_token_count": 653.7375, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.991917508840561, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008082496747374535, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.04449768066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.09104919433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21820, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19944573789834977, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.12591580003499986, + "grad_norm": 0.053689949214458466, + "kept_token_count": 654.8765625, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922827064990998, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077172876335680485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7936218261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63372802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21830, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.1998318538069725, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.12505810633301734, + "grad_norm": 0.04672848433256149, + "kept_token_count": 654.5765625, + "learning_rate": 5.671125561561713e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920318245887756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007968165911734104, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.09772033691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.04515380859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21840, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20027289688587188, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.1284395143389702, + "grad_norm": 0.04170618951320648, + "kept_token_count": 654.5375, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925338327884674, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074661606922745705, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.09352416992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.530224609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21850, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20016376972198485, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.12328853830695152, + "grad_norm": 0.0687977522611618, + "kept_token_count": 654.384375, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918178021907806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00818221103399992, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8947448730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4677398681641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21860, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2000277817249298, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.1222800187766552, + "grad_norm": 0.03877343237400055, + "kept_token_count": 654.709375, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922636151313782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007736396323889494, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1052490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2159881591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21870, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.20087039470672607, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.11649721190333366, + "grad_norm": 0.040819816291332245, + "kept_token_count": 653.9046875, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919170379638672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008082962222397327, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0488708496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9256195068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19900918304920195, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.13131047338247298, + "grad_norm": 0.04957764968276024, + "kept_token_count": 655.7, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921784818172454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007821511197835206, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9927185058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0295928955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21890, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.20025723278522492, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.1247960515320301, + "grad_norm": 0.04224804416298866, + "kept_token_count": 654.4671875, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918487787246704, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008151230122894048, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.99537963867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8944427490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21900, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19955393821001052, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.12043681591749192, + "grad_norm": 0.03713595122098923, + "kept_token_count": 654.9828125, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924286901950836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007571317814290523, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.71604614257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.2610260009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21910, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19939297586679458, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.12705165296792983, + "grad_norm": 0.07808969914913177, + "kept_token_count": 655.0921875, + "learning_rate": 5.567906122683439e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919914543628693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00800854042172432, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.35016174316405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.32265014648436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.1998092472553253, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.13005397096276283, + "grad_norm": 0.03912176936864853, + "kept_token_count": 654.8703125, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917242348194122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008275763224810362, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8655670166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9438018798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21930, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.19995252937078475, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.12269142419099807, + "grad_norm": 0.0489874891936779, + "kept_token_count": 654.7265625, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915070295333862, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008492984343320131, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1573547363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6960876464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21940, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20081471353769303, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.12776417061686515, + "grad_norm": 0.04886981472373009, + "kept_token_count": 654.165625, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916665256023407, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008333465829491616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.24640502929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.42694091796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21950, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19950819313526152, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.11738980263471603, + "grad_norm": 0.040311019867658615, + "kept_token_count": 655.0671875, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992210078239441, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007789915800094605, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.65007934570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55820922851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21960, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.20053744167089463, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.12088543847203255, + "grad_norm": 0.04589230939745903, + "kept_token_count": 654.2578125, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922993898391723, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007700616121292114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.38343200683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.62835693359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21970, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19993278980255128, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.13056205883622168, + "grad_norm": 0.04182206839323044, + "kept_token_count": 654.640625, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915186941623688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008481294102966785, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.06385192871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.81634521484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20023764669895172, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.11961322277784348, + "grad_norm": 0.04231216758489609, + "kept_token_count": 654.1859375, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919966697692871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008003316354006529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4709838867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.37867736816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21990, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19972798973321915, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.12493537217378617, + "grad_norm": 0.0419987253844738, + "kept_token_count": 654.7828125, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920299530029297, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007970053050667048, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0733367919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0569580078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22000, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20038921982049943, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.12710706517100334, + "grad_norm": 0.05538512021303177, + "kept_token_count": 654.30625, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915080368518829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008491962496191264, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2451934814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6244689941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22010, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19999324679374694, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.12094644233584403, + "grad_norm": 0.04667309671640396, + "kept_token_count": 654.6234375, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925472259521484, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007452764827758074, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4295288085938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8690673828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22020, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20054567754268646, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.12350905239582062, + "grad_norm": 0.05466267839074135, + "kept_token_count": 654.009375, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924157679080963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007584235910326242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.28717651367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.54088134765624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22030, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.20028901249170303, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.118727096170187, + "grad_norm": 0.061631303280591965, + "kept_token_count": 654.3625, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.991733717918396, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008266283944249154, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2259857177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.28700866699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.1997924417257309, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.1250483848154545, + "grad_norm": 0.051128074526786804, + "kept_token_count": 654.7671875, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923086941242218, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007691300939768553, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0367462158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.82961120605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19956403970718384, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.11420940458774567, + "grad_norm": 0.046836577355861664, + "kept_token_count": 654.8109375, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921525001525879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0078474972397089, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0385345458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.81602783203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22060, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20003275722265243, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.1216452144086361, + "grad_norm": 0.04562404006719589, + "kept_token_count": 654.3859375, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917187869548798, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008281216025352478, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2738311767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.61163330078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22070, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.200381900370121, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.11330848559737206, + "grad_norm": 0.04562556743621826, + "kept_token_count": 654.428125, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918683767318726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008131635561585426, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5327484130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3665405273438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22080, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.1997105211019516, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.11813000664114952, + "grad_norm": 0.06987231969833374, + "kept_token_count": 654.8515625, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923281311988831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00767186488956213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6224700927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0709197998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22090, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1994543567299843, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.11435700580477715, + "grad_norm": 0.04005923494696617, + "kept_token_count": 655.25, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.991817420721054, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008182575087994337, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3935913085937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4501617431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22100, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20012525469064713, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.13306310251355172, + "grad_norm": 0.04730161279439926, + "kept_token_count": 654.7671875, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917896866798401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008210311364382505, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0046661376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.27808837890626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22110, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.19992878586053847, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.1234818622469902, + "grad_norm": 0.04632186144590378, + "kept_token_count": 654.6765625, + "learning_rate": 5.313083970734013e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919608414173127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008039155509322882, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.18833923339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22120, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.19970587491989136, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.12335973605513573, + "grad_norm": 0.07483112066984177, + "kept_token_count": 655.246875, + "learning_rate": 5.300465428726439e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919641196727753, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008035879395902158, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3409881591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0080993652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22130, + "valid_token_count": 818.75625 + }, + { + "actual_token_mask_ratio": 0.2004306584596634, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.12415240854024887, + "grad_norm": 0.04629265144467354, + "kept_token_count": 654.1640625, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921135365962982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007886455673724414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.85943908691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9291595458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22140, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.20012001991271972, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.12204162925481796, + "grad_norm": 0.05106865614652634, + "kept_token_count": 654.3109375, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918849408626557, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008115044794976711, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3420806884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.84266662597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22150, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.19935376197099686, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.13219091594219207, + "grad_norm": 0.035771314054727554, + "kept_token_count": 655.2796875, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.991825807094574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008174194395542145, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6009918212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4149627685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22160, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.1998037204146385, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.1223301649093628, + "grad_norm": 0.049515899270772934, + "kept_token_count": 654.7875, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924010515213013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007598940748721361, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5278381347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.30853576660155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22170, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20066047012805938, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.13208819255232812, + "grad_norm": 0.04236648231744766, + "kept_token_count": 653.8984375, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922793984413147, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007720606960356235, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7633972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5445953369141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22180, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.19977786689996718, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.12550126165151596, + "grad_norm": 0.049873340874910355, + "kept_token_count": 654.634375, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915771961212159, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008422794379293918, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0622192382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0120788574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.2001303628087044, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.12191078588366508, + "grad_norm": 0.050716299563646317, + "kept_token_count": 654.4453125, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921454191207886, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007854577992111445, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0299041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.599560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22200, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19981189370155333, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.13608898520469664, + "grad_norm": 0.044451307505369186, + "kept_token_count": 654.8046875, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914978861808776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00850211102515459, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.71641845703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.00989685058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22210, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.1998193457722664, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.11833126321434975, + "grad_norm": 0.043730564415454865, + "kept_token_count": 654.7640625, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924790918827057, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007520900759845972, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.54831848144534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.56048889160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22220, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19985053837299346, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.11765117198228836, + "grad_norm": 0.04173729568719864, + "kept_token_count": 654.5625, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925497591495513, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007450238801538944, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5426940917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6111755371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.20007099956274033, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.11511937007308007, + "grad_norm": 0.04601249098777771, + "kept_token_count": 654.6765625, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916512131690979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008348788227885961, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1448455810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9329437255859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22240, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20089649707078933, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.1285310849547386, + "grad_norm": 0.0439363531768322, + "kept_token_count": 653.7484375, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928328394889832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007167158834636212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.44691467285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1066436767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22250, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.19953715950250625, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.11754124760627746, + "grad_norm": 0.06562332063913345, + "kept_token_count": 655.109375, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924233734607697, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007576625142246485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0060119628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3334014892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22260, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20075267404317856, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.1263363167643547, + "grad_norm": 0.051541924476623535, + "kept_token_count": 653.8953125, + "learning_rate": 5.125050305018683e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919921398162842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008007865305989981, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9037261962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.62279357910154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22270, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2010924845933914, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.1305117830634117, + "grad_norm": 0.0637829452753067, + "kept_token_count": 653.4703125, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913465082645416, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008653480838984252, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.380810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7496337890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22280, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.20056856870651246, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.12002479806542396, + "grad_norm": 0.0427272655069828, + "kept_token_count": 654.2171875, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918780565261841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008121939655393363, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6037292480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9788299560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22290, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19977931380271913, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.12868359312415123, + "grad_norm": 0.05863165855407715, + "kept_token_count": 654.69375, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915615797042847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008438412938266993, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7319305419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2046813964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22300, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.1997775539755821, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.12059597596526146, + "grad_norm": 0.04611634835600853, + "kept_token_count": 654.853125, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918949067592621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008105084486305713, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.94605407714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.09210815429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19995446354150773, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.1259261853992939, + "grad_norm": 0.05273151397705078, + "kept_token_count": 654.575, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915744185447692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008425575587898493, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7387298583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3610778808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22320, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20030211359262468, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.12675393223762513, + "grad_norm": 0.04531637951731682, + "kept_token_count": 654.571875, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914842188358307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008515786845237017, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.77122802734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.455322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22330, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.20037973523139954, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.11964227110147477, + "grad_norm": 0.04402506723999977, + "kept_token_count": 653.9765625, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922360002994537, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007763993553817272, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.01854248046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1840515136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22340, + "valid_token_count": 817.8546875 + }, + { + "actual_token_mask_ratio": 0.20025178492069245, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.12650494277477264, + "grad_norm": 0.05085666850209236, + "kept_token_count": 654.5390625, + "learning_rate": 5.025866325666237e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919707119464874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008029290940612555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4987091064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1629974365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20074131935834885, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.12789321169257165, + "grad_norm": 0.06337656080722809, + "kept_token_count": 653.8953125, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921045124530792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007895487640053035, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9278503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6639343261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22360, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20052419006824493, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.1303306847810745, + "grad_norm": 0.03725455328822136, + "kept_token_count": 654.0484375, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921401917934418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00785979237407446, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7519134521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5803161621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22370, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.2003428727388382, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.12400344759225845, + "grad_norm": 0.05296007916331291, + "kept_token_count": 654.0984375, + "learning_rate": 4.9888718917363e-06, + "loss": 0.009, + "mgd_cos_sim": 0.991006749868393, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008993235509842634, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3877716064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9333862304687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22380, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.20021901428699493, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.13208092302083968, + "grad_norm": 0.04334017261862755, + "kept_token_count": 654.7890625, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9928520202636719, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007147953845560551, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7032867431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8880920410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22390, + "valid_token_count": 818.7109375 + }, + { + "actual_token_mask_ratio": 0.20005931705236435, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.1187839850783348, + "grad_norm": 0.058441706001758575, + "kept_token_count": 654.8078125, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913214206695556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008678577281534672, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.12046508789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.93401794433595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22400, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.2003828227519989, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.1229930117726326, + "grad_norm": 0.05065318942070007, + "kept_token_count": 653.9875, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923943400382995, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007605653908103705, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.950146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1908294677734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22410, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.20002220720052719, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.1353883996605873, + "grad_norm": 0.04720378667116165, + "kept_token_count": 654.9296875, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992267233133316, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007732753455638885, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.5734008789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.72821350097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22420, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.20053843557834625, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.12047174349427223, + "grad_norm": 0.05412673577666283, + "kept_token_count": 654.359375, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917697072029114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008230284508317708, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.18223266601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4733856201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22430, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19981468319892884, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.11746635809540748, + "grad_norm": 0.0534026063978672, + "kept_token_count": 654.7015625, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924633502960205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075366456992924215, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4381072998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7006317138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22440, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20013925135135652, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.12437607496976852, + "grad_norm": 0.034455690532922745, + "kept_token_count": 654.428125, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922267377376557, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00777325602248311, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.08221740722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3088134765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22450, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.1996365338563919, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.12368975877761841, + "grad_norm": 0.04781987890601158, + "kept_token_count": 654.934375, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927358806133271, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007264123950153589, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.199755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.85914001464846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19971315562725067, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.13209582343697548, + "grad_norm": 0.045944880694150925, + "kept_token_count": 654.7375, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921122312545776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007887785881757736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.37598571777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.40769958496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22470, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19982143938541413, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.1249259203672409, + "grad_norm": 0.04809096083045006, + "kept_token_count": 654.75, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921892285346985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007810772024095058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.73401794433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3272705078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22480, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2009166345000267, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.11899578049778939, + "grad_norm": 0.053042348474264145, + "kept_token_count": 653.909375, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920784711837769, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007921521551907063, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5875213623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.47755432128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22490, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.20002993941307068, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.12371252030134201, + "grad_norm": 0.04758424684405327, + "kept_token_count": 654.5640625, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923649489879608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076350455172359945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8052215576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.42640686035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22500, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.19960663020610808, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.13099335134029388, + "grad_norm": 0.04082150384783745, + "kept_token_count": 654.7890625, + "learning_rate": 4.829832951373872e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920083463191987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007991651073098183, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.07334289550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5673492431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22510, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19980245232582092, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.12675445675849914, + "grad_norm": 0.04155591502785683, + "kept_token_count": 654.7875, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919274687767029, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008072536997497081, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8138763427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.04648742675784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22520, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.19977373331785203, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.12838130295276642, + "grad_norm": 0.053047724068164825, + "kept_token_count": 654.803125, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.991783881187439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008216126263141632, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1581665039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5568542480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20059698820114136, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.1357489727437496, + "grad_norm": 0.0528152659535408, + "kept_token_count": 653.765625, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922170877456665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007782916538417339, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0596069335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11362915039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22540, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.2006041184067726, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.1176631711423397, + "grad_norm": 0.047033969312906265, + "kept_token_count": 653.771875, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920616328716279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007938364893198014, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.90447998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.74859619140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22550, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.20068733096122743, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.12420616522431374, + "grad_norm": 0.04351853206753731, + "kept_token_count": 654.0265625, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924213826656342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007578625343739986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.8025817871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.76348571777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22560, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19877267330884935, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.1219960480928421, + "grad_norm": 0.05453372374176979, + "kept_token_count": 655.5421875, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918428361415863, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008157165721058846, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0369567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9340759277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19953889548778533, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.12768415808677674, + "grad_norm": 0.04171350225806236, + "kept_token_count": 655.1828125, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913359761238099, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00866401819512248, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.98626403808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9723388671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22580, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.20037280172109603, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.12086737900972366, + "grad_norm": 0.055082276463508606, + "kept_token_count": 654.1640625, + "learning_rate": 4.732998801074911e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920151770114899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007984818052500486, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.42206420898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.56496276855466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22590, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.1999919295310974, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.12208347246050835, + "grad_norm": 0.05571557581424713, + "kept_token_count": 654.5625, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922467112541199, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077532809227705, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.64348754882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0812194824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22600, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19996187835931778, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.12482173219323159, + "grad_norm": 0.04886704310774803, + "kept_token_count": 654.8625, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912923157215119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008707686606794596, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.838134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0062316894531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22610, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.19883116483688354, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.1239296019077301, + "grad_norm": 0.0623137503862381, + "kept_token_count": 655.4453125, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915137708187103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00848623551428318, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2972778320312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8430694580078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22620, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20049837082624436, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.13750763162970542, + "grad_norm": 0.04837491363286972, + "kept_token_count": 654.390625, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9913911044597625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008608893025666475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5517974853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1561706542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22630, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.19979303628206252, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.11613750904798507, + "grad_norm": 0.07514279335737228, + "kept_token_count": 654.646875, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914697468280792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008530252613127232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8140350341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.51531372070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22640, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.19981158822774886, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.12421587333083153, + "grad_norm": 0.054386939853429794, + "kept_token_count": 654.4234375, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918045163154602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008195502869784831, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.59129028320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8185516357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22650, + "valid_token_count": 817.83125 + }, + { + "actual_token_mask_ratio": 0.20039063990116118, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.12530868053436278, + "grad_norm": 0.04647372290492058, + "kept_token_count": 654.3546875, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.992388653755188, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007611342333257198, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.51548461914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.07874450683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22660, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19970318973064421, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.12682760432362555, + "grad_norm": 0.044774603098630905, + "kept_token_count": 654.9203125, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9912778019905091, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008722191955894232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2483825683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6244171142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22670, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.1997477024793625, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.12638093829154967, + "grad_norm": 0.07531628757715225, + "kept_token_count": 654.865625, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915916562080384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408353384584188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1694305419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32220153808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22680, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20072403252124787, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.13342232406139373, + "grad_norm": 0.06841158866882324, + "kept_token_count": 654.090625, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919098615646362, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008090133219957352, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.88251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.51708068847654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22690, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19937558323144913, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.12338663563132286, + "grad_norm": 0.035115234553813934, + "kept_token_count": 654.9671875, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918242812156677, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008175716362893581, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9754333496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6343627929688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22700, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.2001308649778366, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.12690813094377518, + "grad_norm": 0.04524976387619972, + "kept_token_count": 654.3515625, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.991829788684845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008170218206942081, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1719696044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30991821289064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22710, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.1996322050690651, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.12351144924759865, + "grad_norm": 0.08236745744943619, + "kept_token_count": 655.0640625, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914789915084838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008521002251654863, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2633941650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5829711914063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22720, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.19981804192066194, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.12383376061916351, + "grad_norm": 0.034848686307668686, + "kept_token_count": 654.7671875, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915918350219727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008408167865127325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1966918945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.78883972167966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22730, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2002462387084961, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.13057220801711084, + "grad_norm": 0.05358486995100975, + "kept_token_count": 654.1609375, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9928612470626831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007138758338987828, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7691375732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2235046386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22740, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.20061683654785156, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.12155007943511009, + "grad_norm": 0.06883840262889862, + "kept_token_count": 654.2, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914659440517426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00853405026718974, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.28724365234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.71270141601565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22750, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2006073147058487, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.12898602560162545, + "grad_norm": 0.047733839601278305, + "kept_token_count": 654.003125, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923103094100952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007689686771482229, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.64853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4415618896484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22760, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.1998421773314476, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.11899150907993317, + "grad_norm": 0.036266181617975235, + "kept_token_count": 654.6796875, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922894179821015, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007710588630288839, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2026306152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2322692871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19979041814804077, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.13003874197602272, + "grad_norm": 0.1043667122721672, + "kept_token_count": 654.803125, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913169980049134, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008682993985712529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.96087646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.76824951171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19991406202316284, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.1274290218949318, + "grad_norm": 0.037323515862226486, + "kept_token_count": 654.496875, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.992698585987091, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007301414385437965, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1993927001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9683532714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22790, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19965340197086334, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.11015521362423897, + "grad_norm": 0.041066113859415054, + "kept_token_count": 655.1, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.992728841304779, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007271149754524231, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.018603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.37143859863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22800, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.1996759444475174, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.12617936059832574, + "grad_norm": 0.055533938109874725, + "kept_token_count": 654.8046875, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922180652618409, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077819210477173325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5067626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6141296386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22810, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.19968903660774232, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.11679990291595459, + "grad_norm": 0.05172744020819664, + "kept_token_count": 654.9015625, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920653164386749, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007934675458818674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.85882263183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8362762451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22820, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.1998119205236435, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.11555474549531937, + "grad_norm": 0.03965029865503311, + "kept_token_count": 655.1109375, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925301849842072, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074698119424283504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.53370056152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1503509521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22830, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.19934651255607605, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.13363223671913146, + "grad_norm": 0.06069329008460045, + "kept_token_count": 654.9203125, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915635049343109, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008436482585966586, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.13636474609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9400207519531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22840, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.20004966855049133, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.11684370338916779, + "grad_norm": 0.042536020278930664, + "kept_token_count": 654.6453125, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918004989624023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008199504390358926, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2545867919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.33948974609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19940228909254074, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.11997905001044273, + "grad_norm": 0.03734497353434563, + "kept_token_count": 655.2515625, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920518100261688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007948186341673135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.89563903808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9539001464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22860, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.1991124778985977, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.13689369931817055, + "grad_norm": 0.03932590037584305, + "kept_token_count": 655.0234375, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9915236055850982, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008476385846734047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.38837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6326141357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22870, + "valid_token_count": 817.8671875 + }, + { + "actual_token_mask_ratio": 0.20096212029457092, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.12910076677799226, + "grad_norm": 0.03956031799316406, + "kept_token_count": 653.7640625, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923637092113495, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007636300101876259, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4621154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.95557556152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22880, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20044004917144775, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.1256570488214493, + "grad_norm": 0.05145512521266937, + "kept_token_count": 654.028125, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920718789100647, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007928105164319276, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7776306152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.34633483886716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22890, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.20032199025154113, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.1272528052330017, + "grad_norm": 0.04601973295211792, + "kept_token_count": 654.3265625, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992266571521759, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007733428385108709, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.03040771484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.72755432128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22900, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.20034056007862092, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.11961643770337105, + "grad_norm": 0.056565623730421066, + "kept_token_count": 654.303125, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.992485910654068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007514083757996559, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6624053955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.76172790527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20036644637584686, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.1329055570065975, + "grad_norm": 0.066324383020401, + "kept_token_count": 654.128125, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922319531440735, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007768045738339424, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.02735290527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59443359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22920, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.19993606358766555, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.1242974691092968, + "grad_norm": 0.04285876080393791, + "kept_token_count": 654.7046875, + "learning_rate": 4.330450856943101e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920005619525909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007999440282583236, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5454803466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.24656066894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22930, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.1992899000644684, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.12962499335408212, + "grad_norm": 0.06873812526464462, + "kept_token_count": 655.165625, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992157620191574, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007842371053993703, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6396759033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57257690429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22940, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1998349517583847, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.1265511393547058, + "grad_norm": 0.04778176173567772, + "kept_token_count": 654.7546875, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925698161125183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00743020698428154, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5518432617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7673706054687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20057154297828675, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.1257692225277424, + "grad_norm": 0.05548468604683876, + "kept_token_count": 654.0703125, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922333836555481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00776659781113267, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8502716064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3528533935547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22960, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1999024286866188, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.12772071063518525, + "grad_norm": 0.06879102438688278, + "kept_token_count": 654.7625, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.991748309135437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00825169524177909, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2744598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3749969482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22970, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.1997721552848816, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.12640362679958345, + "grad_norm": 0.042277731001377106, + "kept_token_count": 655.3890625, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918109953403473, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008189004845917225, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.35300903320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7959381103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22980, + "valid_token_count": 819.003125 + }, + { + "actual_token_mask_ratio": 0.20087065994739534, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.13293977230787277, + "grad_norm": 0.05034357309341431, + "kept_token_count": 654.0125, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917976975440979, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008202299941331148, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.68934631347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3055114746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22990, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20080577135086058, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.1200911708176136, + "grad_norm": 0.0499223992228508, + "kept_token_count": 653.9015625, + "learning_rate": 4.249415327007353e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919651448726654, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00803485857322812, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.074169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49400939941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23000, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20047139525413513, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.12329220771789551, + "grad_norm": 0.08124518394470215, + "kept_token_count": 654.121875, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918060839176178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008193918038159609, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.15257263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.43613891601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23010, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20061722993850709, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.11833581626415253, + "grad_norm": 0.04362102597951889, + "kept_token_count": 654.0625, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918458521366119, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008154150936752558, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.07274475097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1636596679688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23020, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.19925552755594253, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.12789359763264657, + "grad_norm": 0.04295247420668602, + "kept_token_count": 655.1203125, + "learning_rate": 4.214881477853227e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919730424880981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008026947546750308, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.91590881347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.99109802246096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23030, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19945300817489625, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.12593360915780066, + "grad_norm": 0.0517013818025589, + "kept_token_count": 655.0890625, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922123789787293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077876266092062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9012939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87636413574216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23040, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20014440417289733, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.1414913073182106, + "grad_norm": 0.050304122269153595, + "kept_token_count": 654.49375, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916727304458618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0083272703923285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3121734619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.36871948242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23050, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.199241741001606, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.12776235565543176, + "grad_norm": 0.04402172192931175, + "kept_token_count": 655.3828125, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919227361679077, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008077269326895476, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5316864013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.381689453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23060, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.1990811049938202, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.11755683273077011, + "grad_norm": 0.05822541192173958, + "kept_token_count": 655.6109375, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0085, + "mgd_cos_sim": 0.9914972901344299, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008502728678286076, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.28351135253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.79267578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23070, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.2003639355301857, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.13029262721538543, + "grad_norm": 0.05294295400381088, + "kept_token_count": 654.4359375, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915577650070191, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008442225400358438, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.66241455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0581481933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23080, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2008599042892456, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.1163999043405056, + "grad_norm": 0.039996787905693054, + "kept_token_count": 653.903125, + "learning_rate": 4.14616798889649e-06, + "loss": 0.008, + "mgd_cos_sim": 0.991992461681366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008007550891488791, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6805908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7326385498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23090, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.1996459648013115, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.12615518718957902, + "grad_norm": 0.05726848170161247, + "kept_token_count": 655.01875, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922292649745941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007770735677331686, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.99120178222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2642395019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23100, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2001517504453659, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.12123484760522843, + "grad_norm": 0.04846598953008652, + "kept_token_count": 654.653125, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921054840087891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007894509006291628, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4982666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2727844238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23110, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.2002324417233467, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.12541958317160606, + "grad_norm": 0.04298032447695732, + "kept_token_count": 654.440625, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921667337417602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007833271101117135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3613006591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.71422424316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23120, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.20077131390571595, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.11933223977684974, + "grad_norm": 0.03768155351281166, + "kept_token_count": 654.0984375, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924416303634643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007558369170874357, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.56565246582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.53090515136716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23130, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20056426376104355, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.12227923944592475, + "grad_norm": 0.05743139982223511, + "kept_token_count": 654.3484375, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917518377304078, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008248167298734187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2753448486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0619201660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23140, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.1998496100306511, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.12298858389258385, + "grad_norm": 0.051268912851810455, + "kept_token_count": 654.89375, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923942446708679, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007605744246393442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1784637451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7658996582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.20019922852516175, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.13509003594517707, + "grad_norm": 0.04520978406071663, + "kept_token_count": 654.4890625, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917901992797852, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008209800347685814, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7655364990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8282775878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23160, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.20047466456890106, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.1233491137623787, + "grad_norm": 0.05421661585569382, + "kept_token_count": 653.8953125, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922710537910462, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007728945557028055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2283630371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7540802001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23170, + "valid_token_count": 817.85625 + }, + { + "actual_token_mask_ratio": 0.19985705912113189, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.12287715077400208, + "grad_norm": 0.04411109536886215, + "kept_token_count": 654.8625, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922345876693726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007765406556427479, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.68414916992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2027557373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23180, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2005041778087616, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.1309356115758419, + "grad_norm": 0.04825832322239876, + "kept_token_count": 654.4890625, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920879185199738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00791207393631339, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4251434326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1200805664063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23190, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.20074676871299743, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.12190619856119156, + "grad_norm": 0.06031787022948265, + "kept_token_count": 654.1625, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919285833835602, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008071428537368775, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4652496337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0206359863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20035104602575302, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.12208721265196801, + "grad_norm": 0.054054901003837585, + "kept_token_count": 654.5984375, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923275113105774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007672488689422607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2774261474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6546905517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23210, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.1996491879224777, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.12766010090708732, + "grad_norm": 0.057030703872442245, + "kept_token_count": 654.9046875, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924437463283539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007556253578513861, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8061828613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8569061279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23220, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.1999506890773773, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.12264583930373192, + "grad_norm": 0.04978053271770477, + "kept_token_count": 654.5078125, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922543048858643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007745705451816321, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2770202636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6470886230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23230, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.19991766512393952, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.12907900288701057, + "grad_norm": 0.046734631061553955, + "kept_token_count": 654.85, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919574201107025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00804255399852991, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.29332580566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.56879272460935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23240, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.1994787946343422, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.12766830921173095, + "grad_norm": 0.039853159338235855, + "kept_token_count": 654.975, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0089, + "mgd_cos_sim": 0.9911185324192047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008881482295691967, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9042602539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.97015686035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23250, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19902015030384063, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.13731942176818848, + "grad_norm": 0.0389828160405159, + "kept_token_count": 655.31875, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918192446231842, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00818075006827712, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0842559814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.06005859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23260, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.20115021169185637, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.12840020507574082, + "grad_norm": 0.042245540767908096, + "kept_token_count": 653.659375, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920593976974488, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007940600533038378, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3315734863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3764343261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23270, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19948730766773223, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.12565779834985732, + "grad_norm": 0.041009481996297836, + "kept_token_count": 654.9484375, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.99166299700737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008337006252259017, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.23737182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0837677001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19946956932544707, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.1321142189204693, + "grad_norm": 0.04877348244190216, + "kept_token_count": 655.25, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917445600032806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008255446515977382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1154754638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.20233764648435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23290, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.19923606514930725, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.11516028419137, + "grad_norm": 0.04833650961518288, + "kept_token_count": 655.2703125, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925596177577972, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007440374232828617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.72295837402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22235107421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23300, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2002000167965889, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.1266752928495407, + "grad_norm": 0.0354655459523201, + "kept_token_count": 654.5328125, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921619415283203, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007838056702166796, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0223724365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.045458984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23310, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20053269267082213, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.11964978650212288, + "grad_norm": 0.04654020071029663, + "kept_token_count": 654.3171875, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925233423709869, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007476662937551737, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0623748779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5985473632812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23320, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.20025731921195983, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.12108329460024833, + "grad_norm": 0.04504501447081566, + "kept_token_count": 654.3984375, + "learning_rate": 3.876091596002746e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920498371124268, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00795015338808298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.93321533203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5862548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23330, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19971898198127747, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.12971656695008277, + "grad_norm": 0.04508638009428978, + "kept_token_count": 654.7609375, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920822322368622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007917758263647556, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.95655822753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6883026123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23340, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20047001540660858, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.12405682802200317, + "grad_norm": 0.05911482870578766, + "kept_token_count": 654.190625, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921208024024963, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007879212591797113, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2563446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.75169372558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1995703622698784, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.12358376681804657, + "grad_norm": 0.037169285118579865, + "kept_token_count": 654.7375, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919458091259002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008054188918322325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4177490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0719390869141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23360, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.20029256939888002, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.12889122739434242, + "grad_norm": 0.0536486878991127, + "kept_token_count": 654.340625, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.9914079368114471, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00859207920730114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.93964233398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94566345214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23370, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.19979776442050934, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.12942283153533934, + "grad_norm": 0.04997644200921059, + "kept_token_count": 654.853125, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925430953502655, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007456901017576456, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0204315185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.068408203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23380, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20112680345773698, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.1308017924427986, + "grad_norm": 0.0464835986495018, + "kept_token_count": 653.4296875, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915629923343658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008437026012688875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7324157714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5043090820312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23390, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.2002071723341942, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.13246357217431068, + "grad_norm": 0.042091257870197296, + "kept_token_count": 654.4671875, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925271809101105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007472825981676578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.46144409179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7522338867187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23400, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20073416978120803, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.11789943128824235, + "grad_norm": 0.03786833956837654, + "kept_token_count": 653.996875, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921496510505676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00785034205764532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9439636230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6621032714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.2001044660806656, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.12564533203840256, + "grad_norm": 0.041328493505716324, + "kept_token_count": 654.7421875, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.992605447769165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00739456620067358, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9050720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.32701416015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23420, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.20012241303920747, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.12389183491468429, + "grad_norm": 0.034984465688467026, + "kept_token_count": 654.346875, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921029567718506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007897046394646168, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.90476684570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7729827880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23430, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.20061576515436172, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.11970482617616654, + "grad_norm": 0.0554133877158165, + "kept_token_count": 653.85, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926241040229797, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007375900074839592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.60337829589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.98488159179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23440, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20012334138154983, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.11420662403106689, + "grad_norm": 0.051988594233989716, + "kept_token_count": 654.6453125, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918059766292572, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008194016013294458, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.87083129882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1429809570312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23450, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.199243226647377, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.12232123911380768, + "grad_norm": 0.04527439922094345, + "kept_token_count": 655.1109375, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921415686607361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007858417928218842, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.53993530273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.28167724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23460, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2005179762840271, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.12586884051561356, + "grad_norm": 0.03647268936038017, + "kept_token_count": 654.034375, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925750970840455, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007424901332706213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.50372924804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2748962402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23470, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.19957114160060882, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.13399144932627677, + "grad_norm": 0.038963329046964645, + "kept_token_count": 654.9875, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918190538883209, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008180957566946746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4294799804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7521545410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1994658038020134, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.12424757331609726, + "grad_norm": 0.05065372213721275, + "kept_token_count": 655.0453125, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919283330440521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008071662858128548, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.73570861816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.03370971679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23490, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19908861815929413, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.12268906533718109, + "grad_norm": 0.03434489294886589, + "kept_token_count": 655.2328125, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923439681529999, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007656035013496876, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.80665893554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.06097106933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23500, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.20018779784440993, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.11982159540057183, + "grad_norm": 0.04048014059662819, + "kept_token_count": 654.40625, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926310837268829, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007368915248662233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5782012939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0998138427734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23510, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19992760866880416, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.1307087540626526, + "grad_norm": 0.06936504691839218, + "kept_token_count": 654.8265625, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.992560225725174, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007439772877842188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6088348388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9582794189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23520, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.1995903342962265, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.12033132165670395, + "grad_norm": 0.03660357743501663, + "kept_token_count": 655.14375, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921508967876435, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007849102467298507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.59583740234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3650146484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23530, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20038436204195023, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.13185752853751181, + "grad_norm": 0.03634187579154968, + "kept_token_count": 654.690625, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925591707229614, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007440835516899824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.068310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8052947998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23540, + "valid_token_count": 818.759375 + }, + { + "actual_token_mask_ratio": 0.20021102130413054, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.12806661576032638, + "grad_norm": 0.03671703115105629, + "kept_token_count": 654.4984375, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924281656742096, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075718343257904054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6621124267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.35060424804686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23550, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19956441223621368, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.12782379165291785, + "grad_norm": 0.05581672489643097, + "kept_token_count": 654.725, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923643469810486, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007635636907070875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7614990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.27158508300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23560, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.19981499761343002, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.11701458767056465, + "grad_norm": 0.0656096488237381, + "kept_token_count": 654.9859375, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924185693264007, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007581437192857266, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.17262268066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4530029296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23570, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19943420886993407, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.1225893072783947, + "grad_norm": 0.06303001940250397, + "kept_token_count": 655.1921875, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925277054309845, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007472297456115484, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1266662597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5381591796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23580, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19918412566184998, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.1235581837594509, + "grad_norm": 0.044710222631692886, + "kept_token_count": 655.3625, + "learning_rate": 3.592299907806881e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920351266860962, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007964866235852242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5567199707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1072021484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23590, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.2000892713665962, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.12242501601576805, + "grad_norm": 0.048468004912137985, + "kept_token_count": 654.575, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992292857170105, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00770711749792099, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8409423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.88628540039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23600, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19975012093782424, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.11936131194233894, + "grad_norm": 0.04716823995113373, + "kept_token_count": 654.646875, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922576725482941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077423200942575935, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.35805358886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0080139160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23610, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.19970775544643402, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.12450731992721557, + "grad_norm": 0.044907018542289734, + "kept_token_count": 654.784375, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923226296901703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076773867942392824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3086364746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8071044921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23620, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.20038517862558364, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.11660004779696465, + "grad_norm": 0.05333956703543663, + "kept_token_count": 654.5125, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925533711910248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007446621730923653, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8223083496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.73561096191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23630, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.19942785799503326, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.12262901738286018, + "grad_norm": 0.050071489065885544, + "kept_token_count": 655.0765625, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923668742179871, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076331283897161485, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3046173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8365173339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19921725243330002, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.12394133284687996, + "grad_norm": 0.05599548667669296, + "kept_token_count": 655.1953125, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924073696136475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007592630200088024, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1583618164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2235992431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23650, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2010647252202034, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.12931175976991655, + "grad_norm": 0.04529977962374687, + "kept_token_count": 653.6828125, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928019285202027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007198082655668259, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5941680908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.12859497070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23660, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.1992845430970192, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.12872882857918738, + "grad_norm": 0.03964958339929581, + "kept_token_count": 655.303125, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923676013946533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007632394228130579, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0221008300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8869689941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23670, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19951072633266448, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.12370193973183632, + "grad_norm": 0.07404133677482605, + "kept_token_count": 654.925, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9915673553943634, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008432641718536615, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2243194580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.58872375488284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23680, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.19991172552108766, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.12075918987393379, + "grad_norm": 0.041009627282619476, + "kept_token_count": 654.8015625, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923824012279511, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007617610413581133, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1665893554688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.89317932128904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23690, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19936379045248032, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.1268102280795574, + "grad_norm": 0.06741667538881302, + "kept_token_count": 654.953125, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922215640544891, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077784392982721325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.61339111328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6846893310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23700, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.19987322986125947, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.12899063527584076, + "grad_norm": 0.05075549706816673, + "kept_token_count": 654.5296875, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926608502864838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007339148782193661, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8057556152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.13140869140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23710, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19984719306230544, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.13044366911053656, + "grad_norm": 0.03958336263895035, + "kept_token_count": 654.7375, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921202778816223, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007879716344177723, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.6100769042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.07549743652345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23720, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.1993349313735962, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.12358177229762077, + "grad_norm": 0.04596567898988724, + "kept_token_count": 655.0890625, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922015488147735, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077984720468521115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.99852294921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3466857910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23730, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19970229864120484, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.1266193114221096, + "grad_norm": 0.04487182945013046, + "kept_token_count": 654.98125, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924730360507965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007526966463774443, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0244598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4522216796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23740, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19989493638277053, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.12275771647691727, + "grad_norm": 0.04299332946538925, + "kept_token_count": 654.9203125, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919869899749756, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008013009186834097, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5948181152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0677154541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23750, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.20080789774656296, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.12422619760036469, + "grad_norm": 0.04192602261900902, + "kept_token_count": 653.95, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923803269863128, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007619680184870958, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.018798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.01409912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19982071071863175, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.12799017056822776, + "grad_norm": 0.06576184928417206, + "kept_token_count": 654.725, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0086, + "mgd_cos_sim": 0.991441935300827, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008558046072721481, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.71080322265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6140838623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23770, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.1993621915578842, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.12763346061110498, + "grad_norm": 0.05476582050323486, + "kept_token_count": 655.1296875, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926491498947143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00735086128115654, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.83160705566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5305145263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23780, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20015345960855485, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.12849324122071265, + "grad_norm": 0.040121760219335556, + "kept_token_count": 654.378125, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918152809143066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008184730168431997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.45762939453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.20148010253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23790, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19888510853052138, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.1250235080718994, + "grad_norm": 0.036750901490449905, + "kept_token_count": 655.3578125, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922333538532258, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007766649592667818, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8945617675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1580322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23800, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.19969801604747772, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.12926691994071007, + "grad_norm": 0.049780577421188354, + "kept_token_count": 654.9921875, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920179128646851, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00798209197819233, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.10545349121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31525268554685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23810, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20092667490243912, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.1283416971564293, + "grad_norm": 0.05402132496237755, + "kept_token_count": 653.7609375, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917379796504975, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00826202230527997, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.69219970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9813598632812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23820, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.19928879141807557, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.12594351023435593, + "grad_norm": 0.0534868948161602, + "kept_token_count": 655.2171875, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921478807926178, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007852126751095057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2766967773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3822845458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23830, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2008292019367218, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.11744388341903686, + "grad_norm": 0.039010968059301376, + "kept_token_count": 654.0953125, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919113099575043, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00808868147432804, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.05128173828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6505645751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23840, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.19973034709692, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.13548621088266372, + "grad_norm": 0.04593977332115173, + "kept_token_count": 654.828125, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917125761508941, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00828741081058979, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.04417724609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9515777587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23850, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2001711219549179, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.12006032615900039, + "grad_norm": 0.0561271533370018, + "kept_token_count": 654.3640625, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922688961029053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007731100637465715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3561981201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.20001220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23860, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.20034812688827514, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.1326141342520714, + "grad_norm": 0.04281243309378624, + "kept_token_count": 654.48125, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925716936588287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007428316585719586, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.83384704589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.77931213378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23870, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.19986776560544967, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.1139918863773346, + "grad_norm": 0.036676306277513504, + "kept_token_count": 654.7359375, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922606408596039, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007739369664341212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2986206054687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5099151611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23880, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.20114247798919677, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.1178849495947361, + "grad_norm": 0.04933832958340645, + "kept_token_count": 653.9234375, + "learning_rate": 3.276505234650362e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919892132282258, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00801077326759696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7954956054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3633514404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23890, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.20129023641347885, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.14076925218105316, + "grad_norm": 0.04168925806879997, + "kept_token_count": 653.56875, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927491545677185, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007250845804810524, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.67225646972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5003601074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23900, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.20028214901685715, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.1190324641764164, + "grad_norm": 0.03736811503767967, + "kept_token_count": 654.6125, + "learning_rate": 3.255904319293562e-06, + "loss": 0.007, + "mgd_cos_sim": 0.992976850271225, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007023141253739595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5207061767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.56532592773436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23910, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.20005649626255034, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.11573194265365601, + "grad_norm": 0.04801705479621887, + "kept_token_count": 654.859375, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924161791801452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007583821844309569, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.45103759765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6662658691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23920, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.20031901597976684, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.13275400102138518, + "grad_norm": 0.04115338623523712, + "kept_token_count": 654.4296875, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919358432292938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008064147736877202, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.87559814453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.13873291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23930, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.19965208619832991, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.12511086836457253, + "grad_norm": 0.04532330855727196, + "kept_token_count": 655.0359375, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923182547092437, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007681746780872345, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1887573242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7157928466797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23940, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20022147446870803, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.1290251538157463, + "grad_norm": 0.051167357712984085, + "kept_token_count": 654.690625, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921014726161956, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007898527663201094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9949493408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2754760742188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23950, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19965552389621735, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.1255899600684643, + "grad_norm": 0.034563325345516205, + "kept_token_count": 655.1734375, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923227548599243, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007677235174924135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6103942871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.20302429199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23960, + "valid_token_count": 818.61875 + }, + { + "actual_token_mask_ratio": 0.20108712017536162, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.12624768912792206, + "grad_norm": 0.04769054800271988, + "kept_token_count": 653.6453125, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919015645980835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008098428789526223, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.18422241210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1224670410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19967946857213975, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.12066759169101715, + "grad_norm": 0.031109074130654335, + "kept_token_count": 654.85625, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924791693687439, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00752082271501422, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4660278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1022888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23980, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.20014333575963975, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.11379914954304696, + "grad_norm": 0.043870262801647186, + "kept_token_count": 654.6015625, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922867953777313, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007713205646723509, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.58168029785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.103564453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2003094419836998, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.11703085526823997, + "grad_norm": 0.054616186767816544, + "kept_token_count": 654.3578125, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916647613048554, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008335239533334971, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0695831298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6311309814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24000, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2004973843693733, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.13276746720075608, + "grad_norm": 0.0540279783308506, + "kept_token_count": 653.9828125, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921999871730804, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007800023723393679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5267303466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.75989990234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24010, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.19979678690433503, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.12991334050893782, + "grad_norm": 0.038979001343250275, + "kept_token_count": 654.8765625, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922282040119171, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007771788723766803, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.52979736328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.64329833984374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24020, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2006321907043457, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.12913573682308196, + "grad_norm": 0.0700603649020195, + "kept_token_count": 654.203125, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921032071113587, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007896788138896226, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.89954833984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3195404052734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24030, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19988477379083633, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.11241447851061821, + "grad_norm": 0.03223834186792374, + "kept_token_count": 654.61875, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924791991710663, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007520799804478884, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4572357177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1224060058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24040, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.199497689306736, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.12345576286315918, + "grad_norm": 0.05733500048518181, + "kept_token_count": 655.096875, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917637467384338, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00823625000193715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.44642944335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4986267089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24050, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19994175136089326, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.1240103654563427, + "grad_norm": 0.04770627245306969, + "kept_token_count": 654.4625, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923098027706146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007690200116485357, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7073974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25892639160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24060, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20043281614780425, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.12238066270947456, + "grad_norm": 0.053487379103899, + "kept_token_count": 654.2453125, + "learning_rate": 3.09316025464495e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920103430747986, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007989655714482069, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.21654357910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5514801025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24070, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19986145347356796, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.12101267501711846, + "grad_norm": 0.05127318575978279, + "kept_token_count": 654.8875, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927337765693665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007266227062791586, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.29888305664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7673645019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19996067881584167, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.12613120377063752, + "grad_norm": 0.03933911398053169, + "kept_token_count": 654.41875, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921579658985138, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007842020317912101, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1020172119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8024169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24090, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.2005097508430481, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.12541372776031495, + "grad_norm": 0.03971046581864357, + "kept_token_count": 653.9203125, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926085948944092, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073913936503231525, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4201599121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0358825683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24100, + "valid_token_count": 817.921875 + }, + { + "actual_token_mask_ratio": 0.20043086409568786, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.12342332750558853, + "grad_norm": 0.0347723513841629, + "kept_token_count": 654.3640625, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925226688385009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074773370288312435, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.36781921386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.54755859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24110, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.20150331854820253, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.12037543430924416, + "grad_norm": 0.0678829476237297, + "kept_token_count": 653.115625, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916565358638764, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008343454450368881, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9125701904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.19178161621096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24120, + "valid_token_count": 817.93125 + }, + { + "actual_token_mask_ratio": 0.19976022988557815, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.1312025360763073, + "grad_norm": 0.04515531659126282, + "kept_token_count": 654.728125, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924039661884307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00759604386985302, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5290496826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3151092529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24130, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19980326145887375, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.12428241744637489, + "grad_norm": 0.0576503649353981, + "kept_token_count": 654.7890625, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921132802963257, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007886720634996892, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9494262695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.73468017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24140, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20119640976190567, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.12463693246245384, + "grad_norm": 0.052788566797971725, + "kept_token_count": 653.4671875, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920596837997436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007940311264246703, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.86565551757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.800048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24150, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.19920457154512405, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.1280610166490078, + "grad_norm": 0.03807899355888367, + "kept_token_count": 655.0984375, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923492133617401, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076507825404405596, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.77808837890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0137145996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24160, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.1991337224841118, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.1230644203722477, + "grad_norm": 0.060003235936164856, + "kept_token_count": 655.46875, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924585819244385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00754141304641962, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7406860351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.62049560546876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.19985037297010422, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.12893904224038125, + "grad_norm": 0.03850005939602852, + "kept_token_count": 654.646875, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921416342258453, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00785837071016431, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4475494384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.99720458984376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20037122666835785, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.1225163146853447, + "grad_norm": 0.04882757365703583, + "kept_token_count": 654.3671875, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920956909656524, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00790430260822177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.66636352539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5940704345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24190, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.20011375844478607, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.12353715673089027, + "grad_norm": 0.042814288288354874, + "kept_token_count": 654.55, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9913405299186706, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008659461326897145, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.40016479492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6228271484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24200, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.20039586573839188, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.12084577158093453, + "grad_norm": 0.055369045585393906, + "kept_token_count": 654.4859375, + "learning_rate": 2.953795683518659e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919623136520386, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00803768141195178, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.01465759277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.83336181640624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24210, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19991439133882521, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.12806031629443168, + "grad_norm": 0.051656026393175125, + "kept_token_count": 654.675, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918869256973266, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008113086502999068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.08814697265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.362353515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24220, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20035966783761977, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.1229639433324337, + "grad_norm": 0.047861434519290924, + "kept_token_count": 654.115625, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924271702766418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007572839967906475, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.46876831054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2414245605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24230, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.20112377405166626, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.11977276504039765, + "grad_norm": 0.0388975627720356, + "kept_token_count": 653.778125, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925420880317688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007457912806421519, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.98162841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.12493286132815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24240, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2004636988043785, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.12505122050642967, + "grad_norm": 0.04725761339068413, + "kept_token_count": 654.2921875, + "learning_rate": 2.914502731629641e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919842004776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008015789557248354, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8105072021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6268798828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24250, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19981370866298676, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.12549812868237495, + "grad_norm": 0.0360277034342289, + "kept_token_count": 654.740625, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927927374839782, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007207267079502344, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9926727294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7559814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24260, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.1997850477695465, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.12298048064112663, + "grad_norm": 0.03951729089021683, + "kept_token_count": 655.0046875, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922972559928894, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007702740281820298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.11150817871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1250335693359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24270, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.20044369846582413, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.1236648865044117, + "grad_norm": 0.03909502923488617, + "kept_token_count": 654.4609375, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927336573600769, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007266335189342499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3692565917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8861145019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24280, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.2004539430141449, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.1345636159181595, + "grad_norm": 0.04529998078942299, + "kept_token_count": 654.2828125, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.99175825715065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008241745550185442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.21148986816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.88111267089846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24290, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.19951957762241362, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.12347704470157624, + "grad_norm": 0.03422221168875694, + "kept_token_count": 654.8796875, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925988435745239, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00740116061642766, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.354296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.5061492919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24300, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.20128535777330397, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.12732203379273416, + "grad_norm": 0.050329577177762985, + "kept_token_count": 653.559375, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927541792392731, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00724581740796566, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7232727050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49005432128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24310, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20006501525640488, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.12954909205436707, + "grad_norm": 0.1536065638065338, + "kept_token_count": 654.3890625, + "learning_rate": 2.846305895153472e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919675469398499, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008032457530498504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2488586425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.98578491210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24320, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.19973474144935607, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.11846260949969292, + "grad_norm": 0.04064267873764038, + "kept_token_count": 654.5203125, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926271855831146, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007372811064124107, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1567138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.28760681152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24330, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.20053320229053498, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.12767928168177606, + "grad_norm": 0.04013485461473465, + "kept_token_count": 654.159375, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928276896476745, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007172305509448052, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.51279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3880157470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24340, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20141155123710633, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.12140457481145858, + "grad_norm": 0.06532233953475952, + "kept_token_count": 653.4765625, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918787062168122, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008121297787874937, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6109100341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5972198486328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24350, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20079605132341385, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.12859268561005593, + "grad_norm": 0.047150641679763794, + "kept_token_count": 653.9421875, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9929216504096985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007078361045569181, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.52239685058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0411682128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20085942894220352, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.1251594305038452, + "grad_norm": 0.03964856266975403, + "kept_token_count": 653.7921875, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923262774944306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007673713937401771, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.30042419433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8242645263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24370, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.2001908838748932, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.12804030776023864, + "grad_norm": 0.051667243242263794, + "kept_token_count": 654.2984375, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924225091934205, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007577491272240877, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1649108886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.32969055175784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24380, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19953120350837708, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.12787594124674798, + "grad_norm": 0.07159582525491714, + "kept_token_count": 655.13125, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922729432582855, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007727048639208078, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6090545654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.69408264160154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24390, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2000663623213768, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.12324197217822075, + "grad_norm": 0.05745173618197441, + "kept_token_count": 654.49375, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.99219029545784, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007809698116034269, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5899200439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.86354370117186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24400, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19977385252714158, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.13595043867826462, + "grad_norm": 0.04655458405613899, + "kept_token_count": 654.7078125, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922696888446808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007730304542928934, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.890380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5949005126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24410, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20018693059682846, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.11842608079314232, + "grad_norm": 0.06269512325525284, + "kept_token_count": 654.2265625, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921023547649384, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00789764653891325, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.31947631835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14942321777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24420, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.20100993812084197, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.13072337284684182, + "grad_norm": 0.05120566859841347, + "kept_token_count": 653.98125, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921826243400573, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007817384786903858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.84241638183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0602294921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24430, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.20014278441667557, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.12579949125647544, + "grad_norm": 0.04692467674612999, + "kept_token_count": 654.3, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922958612442017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007704145088791847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3486755371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6179138183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24440, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.1996883898973465, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.13539326637983323, + "grad_norm": 0.04513353854417801, + "kept_token_count": 654.971875, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923954784870148, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00760451490059495, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.36664428710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.74939880371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24450, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19981116354465484, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.12204927280545234, + "grad_norm": 0.0426766574382782, + "kept_token_count": 654.4609375, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925894677639008, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007410540618002415, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2768218994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2220733642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24460, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.19930377453565598, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.12503001019358634, + "grad_norm": 0.03964593634009361, + "kept_token_count": 655.0921875, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928068339824676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007193166203796864, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3455291748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8936004638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24470, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19940241277217866, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.13318135738372802, + "grad_norm": 0.03458995372056961, + "kept_token_count": 655.175, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925189435482025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00748106325045228, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.86914978027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7225341796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24480, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20087258666753768, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.12149292603135109, + "grad_norm": 0.03591234236955643, + "kept_token_count": 653.81875, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925877153873444, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007412275485694409, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3230895996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.51240234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.200372676551342, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.12767374143004417, + "grad_norm": 0.05190020799636841, + "kept_token_count": 654.2703125, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922556340694427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007744367886334658, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5144500732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.15308532714846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24500, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.19992511719465256, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.13021734729409218, + "grad_norm": 0.039706163108348846, + "kept_token_count": 654.7640625, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925657868385315, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007434205524623394, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.83302917480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1905212402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24510, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.20079692900180818, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.12198984399437904, + "grad_norm": 0.047932304441928864, + "kept_token_count": 654.1703125, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922452628612518, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077547370456159115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3676483154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.20325622558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24520, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.2004007026553154, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.12106497511267662, + "grad_norm": 0.06113261729478836, + "kept_token_count": 654.1984375, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927713513374329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007228648848831653, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.778955078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10280151367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20073846578598023, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.1368728406727314, + "grad_norm": 0.04902664199471474, + "kept_token_count": 654.2078125, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927129447460175, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007287050411105156, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8840362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4317321777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24540, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19969299733638762, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.11986202225089074, + "grad_norm": 0.03412983566522598, + "kept_token_count": 654.8859375, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922915756702423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077084236778318885, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.76271667480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35753173828124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24550, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20066981315612792, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.13099326118826865, + "grad_norm": 0.06024307757616043, + "kept_token_count": 653.9265625, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922387957572937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007761215418577194, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.03171691894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.145556640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24560, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.1996050238609314, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.1303629420697689, + "grad_norm": 0.047873176634311676, + "kept_token_count": 655.0734375, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928063869476318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007193609420210123, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6983306884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.81969299316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24570, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.1999390795826912, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.11664880588650703, + "grad_norm": 0.0547792986035347, + "kept_token_count": 654.734375, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923862040042877, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007613787706941366, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.06072387695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3144805908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24580, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.20032442212104798, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.11724388673901558, + "grad_norm": 0.06806349754333496, + "kept_token_count": 654.3109375, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9917978167533874, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008202180732041597, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.89356689453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.88478088378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24590, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20004522800445557, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.1265868902206421, + "grad_norm": 0.03842302784323692, + "kept_token_count": 654.65, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926945924758911, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00730541879311204, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3588165283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2991241455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24600, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2000828355550766, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.12358094155788421, + "grad_norm": 0.043483756482601166, + "kept_token_count": 654.584375, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9928551018238068, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007144895941019058, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.606298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2429504394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24610, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.1996379628777504, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.12802172377705573, + "grad_norm": 0.050520021468400955, + "kept_token_count": 654.7390625, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925529778003692, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007447020336985588, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3341979980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63599548339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24620, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20003403425216676, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.12106266394257545, + "grad_norm": 0.03940272703766823, + "kept_token_count": 654.5390625, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.992706173658371, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007293830346316099, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.60686950683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0404418945312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24630, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.200655397772789, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.1135915532708168, + "grad_norm": 0.04455212876200676, + "kept_token_count": 654.16875, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926689803600312, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007331003248691559, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3597808837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9825805664062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24640, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20044223368167877, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.112246935069561, + "grad_norm": 0.04593773931264877, + "kept_token_count": 653.96875, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927681148052215, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007231881190091371, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4474639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8447509765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24650, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.19925652593374252, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.13340438529849052, + "grad_norm": 0.05208071321249008, + "kept_token_count": 655.2234375, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925145149230957, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007485501375049353, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3694763183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.37864685058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19971883445978164, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.12050126791000366, + "grad_norm": 0.04216156527400017, + "kept_token_count": 654.775, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.992488545179367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007511451095342636, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.66916809082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.07350463867186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24670, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.2001044660806656, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.11928077712655068, + "grad_norm": 0.058739859610795975, + "kept_token_count": 654.4, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922945141792298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077054713852703575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4581329345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.28064880371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24680, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.20006469637155533, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.1316017650067806, + "grad_norm": 0.04413386434316635, + "kept_token_count": 654.8078125, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918299317359924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00817006714642048, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7731597900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.90595703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24690, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.19987690299749375, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.12341428771615029, + "grad_norm": 0.04612930491566658, + "kept_token_count": 654.6109375, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925622463226318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074377541430294515, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.19776306152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6162414550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24700, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.20077751576900482, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.13056031242012978, + "grad_norm": 0.0512816421687603, + "kept_token_count": 653.790625, + "learning_rate": 2.479709800807915e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919981956481934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008001806307584047, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.05649719238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6721160888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24710, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.19985950291156768, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.1278798386454582, + "grad_norm": 0.039630040526390076, + "kept_token_count": 654.5875, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920325517654419, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007967442739754915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7665100097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2610778808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.20003124475479125, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.11493295207619666, + "grad_norm": 0.05414393171668053, + "kept_token_count": 654.6640625, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921363353729248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007863671239465474, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4250885009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.275927734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24730, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2000935584306717, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.13087373077869416, + "grad_norm": 0.045602232217788696, + "kept_token_count": 654.83125, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924559831619263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007544022146612405, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2151580810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6990173339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24740, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.19920466393232344, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.11955131590366364, + "grad_norm": 0.041494209319353104, + "kept_token_count": 655.621875, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919492125511169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008050795178860426, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.29010620117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4179931640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24750, + "valid_token_count": 818.7125 + }, + { + "actual_token_mask_ratio": 0.2001277282834053, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.12774017602205276, + "grad_norm": 0.04834303632378578, + "kept_token_count": 654.31875, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.99222452044487, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007775485794991255, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6836273193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3023376464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24760, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.1997072473168373, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.12739424556493759, + "grad_norm": 0.06042996421456337, + "kept_token_count": 655.0015625, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923483550548553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076516327448189255, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.65001220703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.64191284179685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24770, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20004352927207947, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.11643637418746948, + "grad_norm": 0.05386744439601898, + "kept_token_count": 654.7765625, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926640927791596, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007335898745805025, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1054168701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24780, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.20003554672002793, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.12414094656705857, + "grad_norm": 0.04791606590151787, + "kept_token_count": 654.6953125, + "learning_rate": 2.407348847176735e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919550120830536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008044993039220572, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2147552490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11435546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24790, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20007292330265045, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.12978833615779878, + "grad_norm": 0.04467036947607994, + "kept_token_count": 654.6421875, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922953248023987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007704656943678856, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4063446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7021911621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24800, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.200238461792469, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.12454114556312561, + "grad_norm": 0.05629938095808029, + "kept_token_count": 654.5703125, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926229000091553, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073771046474575995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1115356445313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.594482421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24810, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.20071070343255998, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.1236254744231701, + "grad_norm": 0.06320066004991531, + "kept_token_count": 654.203125, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922639966011048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007735994271934032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.34168701171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53855895996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24820, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19916654229164124, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.1306944638490677, + "grad_norm": 0.040350060909986496, + "kept_token_count": 655.1875, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924745619297027, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007525435648858547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.67637329101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9889343261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24830, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19939323216676713, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.12420080155134201, + "grad_norm": 0.05495528131723404, + "kept_token_count": 655.0046875, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925991415977478, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007400863338261843, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.59991455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3785461425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24840, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19975620210170747, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.12865537852048875, + "grad_norm": 0.050898872315883636, + "kept_token_count": 654.7109375, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926279425621033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007372051291167736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6953918457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.51625366210936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24850, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19931866526603698, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.13245949223637582, + "grad_norm": 0.050930775701999664, + "kept_token_count": 655.4140625, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918703138828278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008129690866917371, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.30071716308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9052307128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24860, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.2007569894194603, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.12448323667049407, + "grad_norm": 0.04125921055674553, + "kept_token_count": 654.1375, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921940445899964, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007805955316871405, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0716888427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4202484130859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24870, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.1995706707239151, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.12561996877193451, + "grad_norm": 0.051862701773643494, + "kept_token_count": 654.9140625, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925408005714417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007459193374961615, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4458892822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5326751708984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24880, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2003689393401146, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.1250320628285408, + "grad_norm": 0.04782179370522499, + "kept_token_count": 654.471875, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921811282634735, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007818883471190929, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.546826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9322509765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24890, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.19971890449523927, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.11247338727116585, + "grad_norm": 0.05090462788939476, + "kept_token_count": 655.0421875, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920580506324768, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007941953837871552, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.870654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67628479003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24900, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19996262192726136, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.12905169799923896, + "grad_norm": 0.05579541251063347, + "kept_token_count": 654.759375, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927804350852967, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007219567801803351, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.26800537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.21336059570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24910, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.1999291256070137, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.13315696865320206, + "grad_norm": 0.044214796274900436, + "kept_token_count": 654.834375, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922393023967743, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007760685775429011, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6311309814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9413818359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24920, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.1997597798705101, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.12526888847351075, + "grad_norm": 0.04730081558227539, + "kept_token_count": 654.971875, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924438774585724, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007556116860359907, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6902221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.04803161621095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24930, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.19930076748132705, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.13176413774490356, + "grad_norm": 0.03476644307374954, + "kept_token_count": 655.034375, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924878716468811, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007512127701193094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2549072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1822875976562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24940, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.20120725333690642, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.11932272166013717, + "grad_norm": 0.04404724761843681, + "kept_token_count": 653.55625, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926429092884064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007357102539390326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1869812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7482666015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24950, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20075104087591172, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.12677421197295188, + "grad_norm": 0.04800046235322952, + "kept_token_count": 653.9609375, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922990739345551, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007700924202799797, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.59955444335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3296325683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24960, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20029937326908112, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.12419980317354203, + "grad_norm": 0.04439697042107582, + "kept_token_count": 654.3296875, + "learning_rate": 2.248125622537373e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919665455818176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008033462520688772, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.80701599121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01584777832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20042881667613982, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.12444227337837219, + "grad_norm": 0.047152355313301086, + "kept_token_count": 654.075, + "learning_rate": 2.239426513680467e-06, + "loss": 0.007, + "mgd_cos_sim": 0.9930095255374909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006990468874573707, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0747833251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.25609436035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24980, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.1998522862792015, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.10682606771588325, + "grad_norm": 0.05149927735328674, + "kept_token_count": 654.5484375, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926767349243164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007323264144361019, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0526458740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1189697265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24990, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.19989978671073913, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.13439923971891404, + "grad_norm": 0.04230372607707977, + "kept_token_count": 654.7921875, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926607251167298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007339286245405674, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0898468017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.18348693847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25000, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.20003712028265, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.12003997936844826, + "grad_norm": 0.0695430338382721, + "kept_token_count": 654.515625, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992334920167923, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076650682836771015, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.14842834472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93231506347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25010, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19825389385223388, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.12194338366389275, + "grad_norm": 0.06625112146139145, + "kept_token_count": 656.015625, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923021256923675, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00769787710160017, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3661712646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9214752197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25020, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2004141166806221, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.1368037350475788, + "grad_norm": 0.03780177980661392, + "kept_token_count": 653.9484375, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9917310118675232, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008268998842686415, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0650238037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1446563720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25030, + "valid_token_count": 817.859375 + }, + { + "actual_token_mask_ratio": 0.20045565515756608, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.1298085018992424, + "grad_norm": 0.05223900452256203, + "kept_token_count": 654.2515625, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924806594848633, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007519326731562614, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5186065673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.402392578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25040, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20042657256126403, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.11700909733772277, + "grad_norm": 0.047696225345134735, + "kept_token_count": 654.1125, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926587224006653, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007341269589960575, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.22206115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.86669006347654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25050, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.19904002845287322, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.13226423859596254, + "grad_norm": 0.04477129876613617, + "kept_token_count": 655.1625, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925029158592225, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007497075758874416, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0923767089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.30638122558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25060, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.19974715113639832, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.12323557436466218, + "grad_norm": 0.07928043603897095, + "kept_token_count": 654.721875, + "learning_rate": 2.161833534674909e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919606626033783, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008039344940334559, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.88727111816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0172454833984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25070, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.20000083446502687, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.13577561005949973, + "grad_norm": 0.044312842190265656, + "kept_token_count": 654.503125, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9929378688335418, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0070621325634419915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.36472778320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.93055725097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19947399348020553, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.13334240019321442, + "grad_norm": 0.04816752299666405, + "kept_token_count": 655.109375, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925832092761994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007416785042732954, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.61353454589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.77752075195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25090, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.1995416060090065, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.12312615215778351, + "grad_norm": 0.08694003522396088, + "kept_token_count": 655.0359375, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992262214422226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007737785484641791, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5157165527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5656463623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25100, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19983769208192825, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.11730113998055458, + "grad_norm": 0.05714628845453262, + "kept_token_count": 654.709375, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925870418548584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007412961404770613, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5851318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.98817749023436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25110, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20069762766361238, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.1200770080089569, + "grad_norm": 0.06548697501420975, + "kept_token_count": 653.9453125, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992317259311676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007682749442756176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.80113220214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.59893188476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25120, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.19980865716934204, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.12955437004566192, + "grad_norm": 0.07859718799591064, + "kept_token_count": 654.6875, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0082, + "mgd_cos_sim": 0.9918333947658539, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0081666000187397, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8322723388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6516845703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25130, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.199046328663826, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.13877770751714708, + "grad_norm": 0.05284198746085167, + "kept_token_count": 655.125, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924173831939698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075826202519237995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.02446899414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6319610595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25140, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.2001892164349556, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.11444041952490806, + "grad_norm": 0.06019281595945358, + "kept_token_count": 654.540625, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924451231956481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007554879691451788, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.55401611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6213806152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25150, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.19978607594966888, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.12699994668364525, + "grad_norm": 0.03566639870405197, + "kept_token_count": 655.1328125, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926961660385132, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007303825579583645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.03392639160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.50970764160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25160, + "valid_token_count": 818.696875 + }, + { + "actual_token_mask_ratio": 0.19985301792621613, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.12390023767948151, + "grad_norm": 0.048932187259197235, + "kept_token_count": 654.50625, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919302523136139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008069746289402246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.3798065185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.54810485839846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.20005109757184983, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.1318957880139351, + "grad_norm": 0.0728151798248291, + "kept_token_count": 654.7640625, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925798535346985, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007420158665627241, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.388232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.05437622070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25180, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20011907070875168, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.12388689443469048, + "grad_norm": 0.055047355592250824, + "kept_token_count": 654.390625, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928205609321594, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007179434131830931, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.55537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.551611328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25190, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.19959262907505035, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.12570858523249626, + "grad_norm": 0.03920658677816391, + "kept_token_count": 655.084375, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921345353126526, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007865462359040976, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.77660217285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6667449951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25200, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20008115470409393, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.13105000406503678, + "grad_norm": 0.08034532517194748, + "kept_token_count": 654.5015625, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.992288064956665, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007711941935122013, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.28289794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7083465576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25210, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20048086941242219, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.12236266061663628, + "grad_norm": 0.04831572249531746, + "kept_token_count": 654.021875, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9927513122558593, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007248687371611595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.02915649414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7516296386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25220, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.200240059196949, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.1181537926197052, + "grad_norm": 0.04695013165473938, + "kept_token_count": 654.184375, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922242581844329, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077757278457283975, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4858703613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.09230346679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25230, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.19897716194391252, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.11927027329802513, + "grad_norm": 0.04221607744693756, + "kept_token_count": 655.3234375, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9929443180561066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007055683992803097, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2334716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.24010009765624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25240, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.2003655031323433, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.12970860451459884, + "grad_norm": 0.034321852028369904, + "kept_token_count": 654.30625, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925273835659028, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007472611218690872, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0718566894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.97303771972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25250, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20011195540428162, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.12676226124167442, + "grad_norm": 0.046449463814496994, + "kept_token_count": 654.275, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920013070106506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007998683024197817, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.1814392089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.69582214355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25260, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19988046139478682, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.12475103512406349, + "grad_norm": 0.0435621440410614, + "kept_token_count": 654.85625, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924689948558807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007531008031219244, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.45179748535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.99110107421876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25270, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.1995774120092392, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.12026897147297859, + "grad_norm": 0.068321593105793, + "kept_token_count": 655.1234375, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919679462909698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00803205305710435, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7051025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.09657287597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25280, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.19938796758651733, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.12955038622021675, + "grad_norm": 0.04934993386268616, + "kept_token_count": 654.7375, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925085067749023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074914941564202305, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.34938659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53016357421876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25290, + "valid_token_count": 817.7984375 + }, + { + "actual_token_mask_ratio": 0.19986699968576432, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.12411201968789101, + "grad_norm": 0.04607327654957771, + "kept_token_count": 654.6546875, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922730565071106, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00772693557664752, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9985107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8634979248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.20042150765657424, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.12998681291937828, + "grad_norm": 0.04435527324676514, + "kept_token_count": 654.55625, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924788296222686, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007521166373044253, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8612396240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.50743713378904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25310, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.1990630954504013, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.12595680728554726, + "grad_norm": 0.045981429517269135, + "kept_token_count": 655.553125, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.007, + "mgd_cos_sim": 0.9930325627326966, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006967450305819511, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4089660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.99965515136716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25320, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.20074552744627, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.13094573989510536, + "grad_norm": 0.04235072061419487, + "kept_token_count": 654.171875, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924650728702545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00753492172807455, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4854736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4944793701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25330, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20027290731668473, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.1277154229581356, + "grad_norm": 0.04254089295864105, + "kept_token_count": 654.0484375, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925002038478852, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007499779388308525, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1268585205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1787841796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25340, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.19923127442598343, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.12945858761668205, + "grad_norm": 0.04797537624835968, + "kept_token_count": 655.1828125, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928087532520294, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007191238552331924, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.18504943847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.45509033203126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19990075826644899, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.12349123731255532, + "grad_norm": 0.05318872258067131, + "kept_token_count": 654.7796875, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.991943109035492, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008056881371885538, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4346954345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8876495361328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25360, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.19943605065345765, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.13152151107788085, + "grad_norm": 0.049721989780664444, + "kept_token_count": 655.0625, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928175330162048, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00718246391043067, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.34332275390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78431701660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25370, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2004857137799263, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.12167366147041321, + "grad_norm": 0.05489574372768402, + "kept_token_count": 654.2359375, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922031462192535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007796855736523866, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.76775512695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53721618652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25380, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19901247173547745, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.12225164696574212, + "grad_norm": 0.046527646481990814, + "kept_token_count": 655.31875, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918723106384277, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008127691131085158, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1432769775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1814025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25390, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.1991357386112213, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.1173960231244564, + "grad_norm": 0.0493573397397995, + "kept_token_count": 655.36875, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924175322055817, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075824717991054055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.93833312988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.61305236816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25400, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19996597468852997, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.13206253722310066, + "grad_norm": 0.039860717952251434, + "kept_token_count": 654.63125, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920668005943298, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007933210395276547, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.56370544433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7047424316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25410, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19998373985290527, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.13857040032744408, + "grad_norm": 0.04332225024700165, + "kept_token_count": 654.934375, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.992742532491684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007257462292909622, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9248779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3808624267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25420, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.20069553107023239, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.12239043191075324, + "grad_norm": 0.061653222888708115, + "kept_token_count": 654.103125, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922741234302521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007725873682647944, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9799102783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.71844787597655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25430, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20010166615247726, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.12385973781347275, + "grad_norm": 0.04713346064090729, + "kept_token_count": 654.43125, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920215487480164, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007978442404419184, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.58628540039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.64806213378904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25440, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.1999814972281456, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.12337831556797027, + "grad_norm": 0.06191492825746536, + "kept_token_count": 654.9921875, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.992361456155777, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007638535462319851, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.11261291503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.90602416992186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25450, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.20026629269123078, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.12279066890478134, + "grad_norm": 0.05421904847025871, + "kept_token_count": 654.4015625, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926595330238343, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00734046921133995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.86240539550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.42723388671874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25460, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20011021047830582, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.12946497201919555, + "grad_norm": 0.038805101066827774, + "kept_token_count": 654.4890625, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923650979995727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007634899392724037, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5978637695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.61020812988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25470, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19931296855211258, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.12559806406497956, + "grad_norm": 0.04108389839529991, + "kept_token_count": 655.06875, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923062324523926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007693776488304138, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.40064392089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5980712890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25480, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19905080050230026, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.12781169191002845, + "grad_norm": 0.050685517489910126, + "kept_token_count": 655.159375, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924130260944366, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007586968597024679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7326385498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.88115844726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25490, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19980616718530655, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.13140224665403366, + "grad_norm": 0.04122387617826462, + "kept_token_count": 654.9296875, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924271881580353, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007572795078158378, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.65271911621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8825988769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20035615116357802, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.12283210307359696, + "grad_norm": 0.05276823043823242, + "kept_token_count": 654.396875, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926339745521545, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007366021536290646, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6638214111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2899139404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25510, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20049024671316146, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.1183314487338066, + "grad_norm": 0.03841446712613106, + "kept_token_count": 654.459375, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.992173182964325, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007826818991452456, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.66041870117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.36626586914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25520, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.20035582929849624, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.12213614955544472, + "grad_norm": 0.05230676010251045, + "kept_token_count": 654.2859375, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924734354019165, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00752656003460288, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2637908935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0132354736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25530, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19970133453607558, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.1301013246178627, + "grad_norm": 0.04550866037607193, + "kept_token_count": 654.9203125, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924869000911712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007513102516531945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.81333312988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4565795898437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25540, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19927124083042144, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.12959547117352485, + "grad_norm": 0.033645208925008774, + "kept_token_count": 655.19375, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921344876289367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007865498680621386, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.726904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0708862304688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25550, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.200907164812088, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.12261927276849746, + "grad_norm": 0.06296101212501526, + "kept_token_count": 654.0421875, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922001540660859, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007799843978136778, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.76987915039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0314117431641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25560, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19972364902496337, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.13040874972939492, + "grad_norm": 0.05765736848115921, + "kept_token_count": 654.79375, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.992462956905365, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007537038903683424, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.62884216308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9973571777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25570, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19995276331901551, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.12129226997494698, + "grad_norm": 0.06375378370285034, + "kept_token_count": 654.8265625, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923297762870789, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076702342368662356, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0916046142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8816619873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25580, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.2002958670258522, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.1340359181165695, + "grad_norm": 0.0414896085858345, + "kept_token_count": 654.390625, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923684537410736, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007631540112197399, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.9996063232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7933013916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25590, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19938427209854126, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.12941506132483482, + "grad_norm": 0.05757521092891693, + "kept_token_count": 655.3890625, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923552334308624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007644755858927965, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.38788452148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.52691040039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25600, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.20041353553533553, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.12903092205524444, + "grad_norm": 0.06851395219564438, + "kept_token_count": 654.1, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921316802501678, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007868331484496593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.16860046386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6930877685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25610, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.20055245012044906, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.12139467969536781, + "grad_norm": 0.03969087824225426, + "kept_token_count": 654.26875, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926392078399658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007360797096043825, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.22850341796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1678161621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20045313835144044, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.13078085705637932, + "grad_norm": 0.04740387573838234, + "kept_token_count": 654.4671875, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926407098770141, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007359294313937425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.99404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5751190185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25630, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19919087886810302, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.11851820051670074, + "grad_norm": 0.04966098070144653, + "kept_token_count": 655.053125, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922211170196533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007778889313340187, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9962615966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.55735168457034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25640, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.19935597479343414, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.13008143231272698, + "grad_norm": 0.04587344825267792, + "kept_token_count": 655.125, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.008, + "mgd_cos_sim": 0.992048579454422, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007951420173048973, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.94879150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8808258056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25650, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20031531751155854, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.11963773742318154, + "grad_norm": 0.04633154347538948, + "kept_token_count": 654.440625, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920085430145263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007991466764360666, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.97084655761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.38995971679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25660, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20044782310724257, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.1296805590391159, + "grad_norm": 0.05762574076652527, + "kept_token_count": 654.2984375, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0072, + "mgd_cos_sim": 0.9928161978721619, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00718380743637681, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7881622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.23833923339845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25670, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19975956678390502, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.12909548133611679, + "grad_norm": 0.04497672989964485, + "kept_token_count": 654.5875, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922465980052948, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007753389980643987, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.54148864746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2187805175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25680, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.20087458491325377, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.12148266732692718, + "grad_norm": 0.04621609300374985, + "kept_token_count": 654.13125, + "learning_rate": 1.662031045573319e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920496225357056, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00795037392526865, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6929443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4325744628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25690, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.19949595332145692, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.12934428378939627, + "grad_norm": 0.06378023326396942, + "kept_token_count": 654.9578125, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925074696540832, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007492535933852196, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.65455932617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.75636291503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25700, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20049106925725937, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.1263808488845825, + "grad_norm": 0.046488963067531586, + "kept_token_count": 653.9109375, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921643018722535, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007835689652711152, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.99510498046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.59181823730466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25710, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.2004401907324791, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.1290664404630661, + "grad_norm": 0.04128175601363182, + "kept_token_count": 654.4484375, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924740970134736, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007525899354368449, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2167633056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.15465393066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25720, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.20060385912656784, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.12004798725247383, + "grad_norm": 0.043469302356243134, + "kept_token_count": 654.009375, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925634503364563, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007436538301408291, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2214752197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4263214111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25730, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19984436482191087, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.12127755880355835, + "grad_norm": 0.058399949222803116, + "kept_token_count": 654.559375, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.99248286485672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00751715237274766, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.72101440429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5133911132813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25740, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2002253755927086, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.12311496883630753, + "grad_norm": 0.04260707646608353, + "kept_token_count": 654.28125, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921727418899536, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007827247492969036, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.7300109863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1097839355469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25750, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.19993235915899277, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.1172540545463562, + "grad_norm": 0.034451816231012344, + "kept_token_count": 654.709375, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923327565193176, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007667235191911459, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.05665893554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2951324462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25760, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.20071317702531816, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.1269024632871151, + "grad_norm": 0.04254765436053276, + "kept_token_count": 654.0125, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921884179115296, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007811595313251018, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.1915740966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0970886230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20008150339126587, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.12467138916254043, + "grad_norm": 0.0443025641143322, + "kept_token_count": 654.4734375, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926819622516632, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073180359788239, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9305694580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.13067932128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25780, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.19869346767663956, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.12984976619482042, + "grad_norm": 0.03985072672367096, + "kept_token_count": 655.5609375, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922124445438385, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007787555828690529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.89515991210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9644805908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25790, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20070952028036118, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.1210184395313263, + "grad_norm": 0.040758099406957626, + "kept_token_count": 653.834375, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921712577342987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007828733790665865, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5968078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2611328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25800, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2002192497253418, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.12823054641485215, + "grad_norm": 0.0455380342900753, + "kept_token_count": 654.48125, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925152957439423, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007484700810164213, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2002288818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.36741027832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25810, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20122261941432953, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.13077810406684875, + "grad_norm": 0.051617056131362915, + "kept_token_count": 653.546875, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925984799861908, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007401510514318943, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.70386657714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8037445068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25820, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.20068545192480086, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.1236199401319027, + "grad_norm": 0.033514603972435, + "kept_token_count": 654.2, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920057058334351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007994290813803673, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8175994873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7647186279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25830, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.20126554518938064, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.12886466905474664, + "grad_norm": 0.06515564769506454, + "kept_token_count": 653.6359375, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921944856643676, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007805506698787212, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1024871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.21087646484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25840, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20013101845979692, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.11372773572802544, + "grad_norm": 0.047330342233181, + "kept_token_count": 654.7390625, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920522689819335, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007947736047208309, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.76361083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.33619079589846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25850, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.19967721551656722, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.12653767690062523, + "grad_norm": 0.0412057600915432, + "kept_token_count": 654.8625, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925374925136566, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007462508697062731, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2539886474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5067199707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25860, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20006594955921173, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.12986110672354698, + "grad_norm": 0.060768429189920425, + "kept_token_count": 654.878125, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918911933898926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00810881108045578, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6625946044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.885888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25870, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.2007313549518585, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.11877615079283714, + "grad_norm": 0.08675163239240646, + "kept_token_count": 654.01875, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919680476188659, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008031951170414687, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.28831176757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.61243896484376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19967684149742126, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.13283251449465752, + "grad_norm": 0.04114483669400215, + "kept_token_count": 654.621875, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925472736358643, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007452720310539007, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3399658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4147247314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25890, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.20007143318653106, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.1321699470281601, + "grad_norm": 0.05654865875840187, + "kept_token_count": 654.45, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923951029777527, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076048864051699635, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7700958251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.853466796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25900, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2000541180372238, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.1270416796207428, + "grad_norm": 0.048232320696115494, + "kept_token_count": 654.4328125, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925061702728272, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007493845280259847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0350311279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.79515380859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25910, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.19954894483089447, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.11900493577122688, + "grad_norm": 0.04414158686995506, + "kept_token_count": 655.20625, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922194182872772, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007780573610216379, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4670837402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25920, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.20018780827522278, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.12238754108548164, + "grad_norm": 0.04292465001344681, + "kept_token_count": 654.209375, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926444828510285, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073555178008973595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.87355651855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2539001464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25930, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.20109915286302565, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.12136917561292648, + "grad_norm": 0.04447849094867706, + "kept_token_count": 653.5390625, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926340520381928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007365959137678147, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.20905151367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.128662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25940, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.1991430103778839, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.12229872196912765, + "grad_norm": 0.043893102556467056, + "kept_token_count": 655.296875, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924084663391113, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007591530214995146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3775238037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0165679931641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25950, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20070902854204178, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.12361387014389039, + "grad_norm": 0.04815185070037842, + "kept_token_count": 654.2890625, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924190819263459, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00758092226460576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3052459716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94256591796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25960, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.20041466802358626, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.1260232761502266, + "grad_norm": 0.05606403574347496, + "kept_token_count": 654.46875, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926131725311279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007386829517781734, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.21795654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.09144287109376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25970, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20056525319814683, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.13726388961076735, + "grad_norm": 0.036047086119651794, + "kept_token_count": 654.303125, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923938870429992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007606111746281385, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7177490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.45880432128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25980, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.19926434457302095, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.13478699401021005, + "grad_norm": 0.0493587888777256, + "kept_token_count": 655.4, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918842136859893, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008115798607468605, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8880584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2071258544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25990, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.1992497503757477, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.126679889857769, + "grad_norm": 0.03182823956012726, + "kept_token_count": 655.046875, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919611990451813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008038808032870293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.806298828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1899932861328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26000, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.20034464150667192, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.13056188449263573, + "grad_norm": 0.05579067021608353, + "kept_token_count": 654.2703125, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0084, + "mgd_cos_sim": 0.9916110694408417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008388921897858382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4491760253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.95896911621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.1997430220246315, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.13231756687164306, + "grad_norm": 0.04387722909450531, + "kept_token_count": 654.771875, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919172763824463, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008082730043679476, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0746337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0135040283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19964611083269118, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.12131896242499352, + "grad_norm": 0.0873977467417717, + "kept_token_count": 655.2703125, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919245064258575, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008075502328574657, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.51259460449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.77113952636716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26030, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.20075661391019822, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.12913054302334787, + "grad_norm": 0.04997171834111214, + "kept_token_count": 653.6796875, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923361659049987, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007663825340569019, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.71555786132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.59754638671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26040, + "valid_token_count": 817.8765625 + }, + { + "actual_token_mask_ratio": 0.19948737174272538, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.12176989763975143, + "grad_norm": 0.045831937342882156, + "kept_token_count": 654.99375, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921798229217529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007820182852447033, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6008544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.03846130371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26050, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.20056271702051162, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.1320300966501236, + "grad_norm": 0.0406193882226944, + "kept_token_count": 654.0578125, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927002668380738, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0072997274808585646, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1542205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0161529541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26060, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.20041385143995286, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.1218427300453186, + "grad_norm": 0.03345092013478279, + "kept_token_count": 654.55, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926925539970398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007307445071637631, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1391326904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5476104736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26070, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.20054872781038285, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.11516607776284218, + "grad_norm": 0.04129611700773239, + "kept_token_count": 653.840625, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925737380981445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074262553825974464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.02642822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8350067138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26080, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.2007839322090149, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.12433510348200798, + "grad_norm": 0.04010888561606407, + "kept_token_count": 654.040625, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923613548278809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007638644892722368, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4954010009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1525421142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26090, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2007604569196701, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.1273054338991642, + "grad_norm": 0.043708521872758865, + "kept_token_count": 654.18125, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923864901065826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007613491546362638, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7492370605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4854339599609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26100, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.19901014417409896, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.13389080464839936, + "grad_norm": 0.04565368592739105, + "kept_token_count": 655.490625, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923459112644195, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007654079422354698, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.63251342773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3195098876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26110, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2012769252061844, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.12776061445474624, + "grad_norm": 0.054085198789834976, + "kept_token_count": 653.534375, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922763407230377, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077236461453139785, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0319091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2562225341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26120, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1991024911403656, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.13159501999616624, + "grad_norm": 0.043363120406866074, + "kept_token_count": 655.2296875, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.99218390583992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007816084753721952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3070129394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3193725585937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26130, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.1993561401963234, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.13072915449738504, + "grad_norm": 0.0529126301407814, + "kept_token_count": 655.0703125, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922290802001953, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007770934514701367, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.4205261230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.51455078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2006314590573311, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.12413157373666764, + "grad_norm": 0.038968101143836975, + "kept_token_count": 654.08125, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921280145645142, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007871991675347089, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6259246826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8281677246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26150, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.200708869099617, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.12058609575033188, + "grad_norm": 0.06342880427837372, + "kept_token_count": 653.740625, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0083, + "mgd_cos_sim": 0.9916531205177307, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00834687715396285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2347076416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.44244384765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26160, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.19927407056093216, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.11661527454853057, + "grad_norm": 0.05396069586277008, + "kept_token_count": 655.5015625, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927480280399322, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00725196236744523, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.14362182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7169647216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26170, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.1989164799451828, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.12277981117367745, + "grad_norm": 0.050195880234241486, + "kept_token_count": 655.26875, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9920652627944946, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007934738509356975, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.48814086914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6451019287109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26180, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.1999173045158386, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.12872374057769775, + "grad_norm": 0.05274369567632675, + "kept_token_count": 654.3, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919791579246521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008020842261612416, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.8124298095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.515234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26190, + "valid_token_count": 817.7875 + }, + { + "actual_token_mask_ratio": 0.1989176094532013, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.12274454310536384, + "grad_norm": 0.061764929443597794, + "kept_token_count": 655.69375, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921330094337464, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00786698078736663, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7050323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.456494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26200, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.1996949791908264, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.12199805527925492, + "grad_norm": 0.039634138345718384, + "kept_token_count": 654.9609375, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924083650112152, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007591638248413801, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.49176635742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8084259033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26210, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19962393641471862, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.12394931316375732, + "grad_norm": 0.05507585033774376, + "kept_token_count": 655.0125, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919418334960938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008058163989335298, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4240325927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.07652893066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26220, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1989678680896759, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.12951621934771537, + "grad_norm": 0.04945920407772064, + "kept_token_count": 655.4125, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921075582504273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00789244519546628, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.12839660644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.06480102539064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19999754875898362, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.11962700709700584, + "grad_norm": 0.04189009219408035, + "kept_token_count": 654.4171875, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927082359790802, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007291771750897169, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1096984863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.79039611816404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26240, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2010029673576355, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.11574585139751434, + "grad_norm": 0.0535685233771801, + "kept_token_count": 653.815625, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925354957580567, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007464517746120692, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.94488830566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.46421508789064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26250, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2004791632294655, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.1318541146814823, + "grad_norm": 0.04486003518104553, + "kept_token_count": 654.2140625, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925739526748657, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007426036149263382, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8011108398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81427612304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26260, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19998535364866257, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.11715183779597282, + "grad_norm": 0.05059453845024109, + "kept_token_count": 654.6828125, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922446250915528, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007755371648818255, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4407592773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.35057678222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20033465921878815, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.12351191267371178, + "grad_norm": 0.051298897713422775, + "kept_token_count": 654.4, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9920047163963318, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007995287980884313, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5655242919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.86301574707034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26280, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1997441455721855, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.1316155456006527, + "grad_norm": 0.036128658801317215, + "kept_token_count": 654.4671875, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925888180732727, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007411180064082146, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2858154296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4182952880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26290, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.20013717710971832, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.11743354350328446, + "grad_norm": 0.05276121199131012, + "kept_token_count": 654.9078125, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.99247527718544, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007524734176695347, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.964404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.915234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26300, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.20034330189228058, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.12429797574877739, + "grad_norm": 0.043419141322374344, + "kept_token_count": 654.59375, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9929472982883454, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007052697613835335, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.31854248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3577606201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26310, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.1997384697198868, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.12094920799136162, + "grad_norm": 0.06353559345006943, + "kept_token_count": 654.8296875, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.008, + "mgd_cos_sim": 0.9919908165931701, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008009184896945954, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.55397644042966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.82273559570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26320, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19999088793992997, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.129940165579319, + "grad_norm": 0.04716847836971283, + "kept_token_count": 655.1046875, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9929087638854981, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007091234344989062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.25584106445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9547485351562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26330, + "valid_token_count": 818.8765625 + }, + { + "actual_token_mask_ratio": 0.19995401948690414, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.13610915765166282, + "grad_norm": 0.04107996076345444, + "kept_token_count": 654.675, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9923273384571075, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007672660704702139, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1382751464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59543762207034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26340, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.20037606358528137, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.12315574064850807, + "grad_norm": 0.043654851615428925, + "kept_token_count": 654.0796875, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.992535126209259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007464879471808672, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6794799804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78440246582034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26350, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.20029893666505813, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.12257200703024865, + "grad_norm": 0.04379991814494133, + "kept_token_count": 654.2640625, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926160573959351, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007383937761187553, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.33550720214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.69969177246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26360, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.1999269336462021, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.13169828355312346, + "grad_norm": 0.06462318450212479, + "kept_token_count": 654.7796875, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924966216087341, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007503367774188518, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5183471679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.51939392089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26370, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19961746782064438, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.12620684877038002, + "grad_norm": 0.052196256816387177, + "kept_token_count": 655.1046875, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9918509542942047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008149054739624262, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.76484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8581146240234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26380, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.20015995502471923, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.13061968982219696, + "grad_norm": 0.048970140516757965, + "kept_token_count": 654.221875, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927104413509369, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007289582770317793, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2486511230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.03028564453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26390, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.19994096606969833, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.12271181344985962, + "grad_norm": 0.04218719154596329, + "kept_token_count": 654.796875, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922937095165253, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007706299982964992, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0032043457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.01678161621095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26400, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20035024285316466, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.11740916892886162, + "grad_norm": 0.049241017550230026, + "kept_token_count": 654.5875, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924784898757935, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075215033255517484, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.209521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6699188232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.20023452043533324, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.12332868501543999, + "grad_norm": 0.05188707634806633, + "kept_token_count": 654.421875, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0077, + "mgd_cos_sim": 0.9922515094280243, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007748495042324066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6526641845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.29253845214845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26420, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19943453073501588, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.1284939840435982, + "grad_norm": 0.04613029584288597, + "kept_token_count": 654.9828125, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9922100067138672, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007790003065019846, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4029968261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1565826416016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26430, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.19926093965768815, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.1222199909389019, + "grad_norm": 0.041407354176044464, + "kept_token_count": 655.4203125, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0071, + "mgd_cos_sim": 0.9928604960441589, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007139500882476568, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7470642089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.80881958007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26440, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19993429034948348, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.12335272878408432, + "grad_norm": 0.0364985354244709, + "kept_token_count": 654.7171875, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.007, + "mgd_cos_sim": 0.9929951429367065, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007004871126264333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.20411682128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7364929199219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26450, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19931021481752395, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.12140877470374108, + "grad_norm": 0.04745233803987503, + "kept_token_count": 655.24375, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926974415779114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007302549760788679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6521514892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.48475341796876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26460, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.20007176101207733, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.12268764153122902, + "grad_norm": 0.051208704710006714, + "kept_token_count": 654.5375, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924002647399902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007599744107574224, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7479187011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0067108154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26470, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.2000396341085434, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.12829467430710792, + "grad_norm": 0.05839679390192032, + "kept_token_count": 654.4734375, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.992483115196228, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007516877911984921, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4140899658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5374237060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26480, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19956284612417222, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.12754333317279815, + "grad_norm": 0.04899017512798309, + "kept_token_count": 655.05, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926211595535278, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007378833740949631, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1304901123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.85886840820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26490, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20012375563383103, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.13556186705827714, + "grad_norm": 0.03961929678916931, + "kept_token_count": 654.3640625, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9924524962902069, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007547508087009191, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9492462158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40452270507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26500, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.19898751080036164, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.1321562871336937, + "grad_norm": 0.0574929416179657, + "kept_token_count": 655.5484375, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927020072937012, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007297991495579481, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.59348754882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.27528076171876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26510, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20070427358150483, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.11619676053524017, + "grad_norm": 0.043671440333127975, + "kept_token_count": 654.2609375, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.007, + "mgd_cos_sim": 0.9929547369480133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007045254576951265, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.88861694335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1073974609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26520, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2006523534655571, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.1251463435590267, + "grad_norm": 0.04157262668013573, + "kept_token_count": 653.8359375, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.007, + "mgd_cos_sim": 0.9930150210857391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006984993722289801, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2440704345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09324951171874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26530, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.2007462352514267, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.12616332471370698, + "grad_norm": 0.03984417766332626, + "kept_token_count": 653.871875, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0079, + "mgd_cos_sim": 0.9921158492565155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00788414916023612, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.564013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.44361572265626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26540, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.19963609278202057, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.12446555942296982, + "grad_norm": 0.05857378989458084, + "kept_token_count": 654.9703125, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924346745014191, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007565327361226082, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.08380126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5530029296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26550, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20004129707813262, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.12094611302018166, + "grad_norm": 0.03787827491760254, + "kept_token_count": 654.6078125, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9923837840557098, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007616215944290161, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.69968566894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.12993774414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26560, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20022662729024887, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.14393532127141953, + "grad_norm": 0.05690914765000343, + "kept_token_count": 654.3640625, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925737142562866, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007426293939352036, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9070739746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9455291748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26570, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19942955374717714, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.12442350536584854, + "grad_norm": 0.053276997059583664, + "kept_token_count": 655.0421875, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9927130043506622, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007286987546831369, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2285583496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.34674682617185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26580, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19905625730752946, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.1286253236234188, + "grad_norm": 0.03626710921525955, + "kept_token_count": 655.30625, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9925755441188813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007424463052302599, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.81011352539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7525939941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26590, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2006703272461891, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.13337482511997223, + "grad_norm": 0.055411446839571, + "kept_token_count": 653.9890625, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0081, + "mgd_cos_sim": 0.9919128835201263, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008087099809199572, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.27366638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.35398559570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26600, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19945354759693146, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.1271384358406067, + "grad_norm": 0.04748634248971939, + "kept_token_count": 654.996875, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0074, + "mgd_cos_sim": 0.9926118969917297, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007388093695044518, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8265045166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.97227172851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26610, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20006236881017686, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.12146876901388168, + "grad_norm": 0.07492097467184067, + "kept_token_count": 654.5203125, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0076, + "mgd_cos_sim": 0.9924001872539521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007599814515560866, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2441711425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.23650512695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26620, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20040459632873536, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.12994546368718146, + "grad_norm": 0.03668782860040665, + "kept_token_count": 654.2578125, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0075, + "mgd_cos_sim": 0.9925015509128571, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007498446200042963, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.91559448242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5864227294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26630, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20008856505155564, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.12928179427981376, + "grad_norm": 0.04632081091403961, + "kept_token_count": 654.7171875, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0069, + "mgd_cos_sim": 0.9930966615676879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006903344858437776, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2168884277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3013946533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26640, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.2000698834657669, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.11831393837928772, + "grad_norm": 0.05237637087702751, + "kept_token_count": 654.5671875, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0078, + "mgd_cos_sim": 0.9921531558036805, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007846830505877733, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5555694580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5475219726562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.20119570195674896, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.12816492319107056, + "grad_norm": 0.03792891651391983, + "kept_token_count": 653.5734375, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0073, + "mgd_cos_sim": 0.9926750123500824, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007324996683746576, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9064697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.87123413085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26660, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19906357526779175, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.12353821769356728, + "grad_norm": 0.05341764912009239, + "kept_token_count": 655.4125, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922215819358826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007778417877852916, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9676208496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8875762939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26670, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.1994676724076271, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.12856118902564048, + "grad_norm": 0.043163884431123734, + "kept_token_count": 655.0484375, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926776051521301, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007322401180863381, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.48716125488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9275848388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26680, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19891190826892852, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.12444354966282845, + "grad_norm": 0.05139876902103424, + "kept_token_count": 655.653125, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923570394515991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007642958126962185, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3919189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0192596435547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26690, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.1997078999876976, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.1311483897268772, + "grad_norm": 0.035114940255880356, + "kept_token_count": 654.9796875, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.992728191614151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007271803822368384, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.55255126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7349914550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26700, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20068297535181046, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.12900199219584466, + "grad_norm": 0.0372718945145607, + "kept_token_count": 654.2203125, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927100241184235, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007289979886263609, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.79204711914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26710, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.2004089504480362, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.11571402028203011, + "grad_norm": 0.06136833503842354, + "kept_token_count": 654.13125, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.9921377301216125, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00786227686330676, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1853363037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.9179962158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.1994973286986351, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.12978054210543633, + "grad_norm": 0.046568527817726135, + "kept_token_count": 654.98125, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923919677734375, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00760803846642375, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0244659423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.68073120117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26730, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.200492762029171, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.12609122842550277, + "grad_norm": 0.040895406156778336, + "kept_token_count": 653.9921875, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0081, + "mgd_cos_sim": 0.9918797969818115, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008120188675820827, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6101379394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.21112365722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26740, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.19980025589466094, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.12704215794801713, + "grad_norm": 0.035877857357263565, + "kept_token_count": 654.9609375, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924866437911988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007513369154185057, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7405700683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.33898010253904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26750, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.19921188205480575, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.12761581540107728, + "grad_norm": 0.03820906952023506, + "kept_token_count": 655.415625, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924219250679016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00757807744666934, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3784698486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.39244384765624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26760, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.20054506957530976, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.1297401636838913, + "grad_norm": 0.036800991743803024, + "kept_token_count": 654.05, + "learning_rate": 9.41346596396453e-07, + "loss": 0.007, + "mgd_cos_sim": 0.992970222234726, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00702977878972888, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7213623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7976501464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26770, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.2003313273191452, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.1223113477230072, + "grad_norm": 0.034813083708286285, + "kept_token_count": 654.5125, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923475682735443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076524392701685425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.02203979492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.79596557617185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26780, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.199234439432621, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.13214274495840073, + "grad_norm": 0.07400868088006973, + "kept_token_count": 655.284375, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922460675239563, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007753932941704988, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7659545898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9026824951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19905351847410202, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.13206373974680902, + "grad_norm": 0.04088699817657471, + "kept_token_count": 655.353125, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.992479395866394, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007520605064928532, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.08432922363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1363586425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26800, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19883789122104645, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.12560857012867926, + "grad_norm": 0.036627255380153656, + "kept_token_count": 655.584375, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.992551577091217, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007448420487344265, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.23874816894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4604034423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20038658678531646, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.12644362673163415, + "grad_norm": 0.06679141521453857, + "kept_token_count": 654.2296875, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923095405101776, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007690460328012705, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2031585693359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6525573730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26820, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19977841675281524, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.1259372889995575, + "grad_norm": 0.04844149947166443, + "kept_token_count": 654.965625, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922394573688507, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00776052176952362, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4467376708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.12782897949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26830, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.20033856630325317, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.11189197897911071, + "grad_norm": 0.04428543522953987, + "kept_token_count": 654.3296875, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9929318845272064, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007068117614835501, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.63169250488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.48948669433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26840, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19928149431943892, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.12956895157694817, + "grad_norm": 0.04217406362295151, + "kept_token_count": 655.2671875, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0083, + "mgd_cos_sim": 0.9916899919509887, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008310016244649887, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5088592529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.74119873046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26850, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19999860227108002, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.13310871198773383, + "grad_norm": 0.054501019418239594, + "kept_token_count": 654.5734375, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926134049892426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007386597339063883, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.9899871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8755767822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1995822623372078, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.11918397322297096, + "grad_norm": 0.05760708451271057, + "kept_token_count": 654.8859375, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925318002700806, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007468210533261299, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.086767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.13983154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26870, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.20110582560300827, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.12409223690629005, + "grad_norm": 0.03920263424515724, + "kept_token_count": 653.9515625, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927782595157624, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007221760880202055, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.46741638183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.34800720214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26880, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.2000372976064682, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.1273101046681404, + "grad_norm": 0.03990491107106209, + "kept_token_count": 654.4640625, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925374984741211, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007462496403604746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2648468017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.46290588378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26890, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20000765174627305, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.12032997235655785, + "grad_norm": 0.030583690851926804, + "kept_token_count": 654.6109375, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925959527492523, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007404055166989565, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9979309082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5838165283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26900, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20014120787382125, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.12402381151914596, + "grad_norm": 0.05551167577505112, + "kept_token_count": 654.559375, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921894907951355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007810507528483867, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.4056121826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7150054931641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26910, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.2001003935933113, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.13045177683234216, + "grad_norm": 0.0406649112701416, + "kept_token_count": 654.4984375, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926810383796691, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007318967673927546, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.72083740234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3090393066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26920, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19931659251451492, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.13139460757374763, + "grad_norm": 0.059596724808216095, + "kept_token_count": 655.140625, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.992736542224884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007263455446809531, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1287780761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.57822265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26930, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.19985697716474532, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.12321617603302001, + "grad_norm": 0.04502058029174805, + "kept_token_count": 654.75625, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927494406700135, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007250546757131815, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1412780761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0470245361328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26940, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20100926160812377, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.12139429301023483, + "grad_norm": 0.038122206926345825, + "kept_token_count": 653.7484375, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0069, + "mgd_cos_sim": 0.9931129157543183, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00688708396628499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7638763427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4232879638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26950, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.200309681892395, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.1292826861143112, + "grad_norm": 0.052235934883356094, + "kept_token_count": 654.2859375, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9922776341438293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007722353376448154, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.74135131835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.83408508300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26960, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2000242739915848, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.12664383575320243, + "grad_norm": 0.053569428622722626, + "kept_token_count": 654.440625, + "learning_rate": 8.294674786719064e-07, + "loss": 0.007, + "mgd_cos_sim": 0.9929820537567139, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007017955090850592, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.2372985839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8487091064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26970, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.20027226060628892, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.129681283980608, + "grad_norm": 0.06095688417553902, + "kept_token_count": 654.4296875, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926214039325714, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007378589827567339, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4996063232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6308197021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26980, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.2007182776927948, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.12213879302144051, + "grad_norm": 0.04450417682528496, + "kept_token_count": 653.884375, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923251986503601, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00767479045316577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0556579589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0766235351563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26990, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.20095354020595552, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.12836465165019034, + "grad_norm": 0.06606689840555191, + "kept_token_count": 653.9640625, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924761950969696, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007523791398853064, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7565093994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.23349609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27000, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.19939475506544113, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.12825927734375, + "grad_norm": 0.03353746980428696, + "kept_token_count": 654.840625, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0068, + "mgd_cos_sim": 0.9931845247745514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006815471593290568, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.33619995117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9438140869141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27010, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.20019025951623917, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.12352434396743775, + "grad_norm": 0.05303315445780754, + "kept_token_count": 654.4484375, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928898930549621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00711011877283454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5277374267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.107861328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27020, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19942562580108641, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.12787231355905532, + "grad_norm": 0.04911774396896362, + "kept_token_count": 655.159375, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926671922206879, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00733279837295413, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7241485595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8569580078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27030, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20040085762739182, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.12536609992384912, + "grad_norm": 0.0526195727288723, + "kept_token_count": 654.5265625, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925192832946778, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007480716239660979, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6873809814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6549499511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27040, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.20016498267650604, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.12350360974669457, + "grad_norm": 0.046065639704465866, + "kept_token_count": 654.7359375, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925590634346009, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074409362860023975, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.00234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5216369628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27050, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.20055584311485292, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.12467941865324975, + "grad_norm": 0.0378139354288578, + "kept_token_count": 654.4234375, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927576303482055, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007242353353649378, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.39707946777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.748291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27060, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.2014099031686783, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.13355302885174752, + "grad_norm": 0.04811876639723778, + "kept_token_count": 653.4390625, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0082, + "mgd_cos_sim": 0.9917901873588562, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008209805749356747, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.36399230957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.119384765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27070, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2003852128982544, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.13766636773943902, + "grad_norm": 0.03876480087637901, + "kept_token_count": 654.325, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925600767135621, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007439938746392727, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.13849487304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0843780517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27080, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.2003970742225647, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.12760531455278395, + "grad_norm": 0.06086672469973564, + "kept_token_count": 654.3, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926535308361053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007346472889184952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3331604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.464111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19996938556432725, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.12085457146167755, + "grad_norm": 0.0392783097922802, + "kept_token_count": 654.484375, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922469615936279, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007753045111894607, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8319061279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.899462890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27100, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.1989728793501854, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.12591043263673782, + "grad_norm": 0.05233987420797348, + "kept_token_count": 655.5078125, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927381932735443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0072618014179170135, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.82709350585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6233856201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20105803906917571, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.1157647892832756, + "grad_norm": 0.04521767795085907, + "kept_token_count": 653.7359375, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927747189998627, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00722529049962759, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7213073730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7134063720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27120, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19975039213895798, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.1209830790758133, + "grad_norm": 0.03306803107261658, + "kept_token_count": 655.165625, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925869941711426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007412999030202627, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.79405212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0644256591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27130, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.200202576816082, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.1253461755812168, + "grad_norm": 0.042898230254650116, + "kept_token_count": 654.4828125, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923308491706848, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007669130060821772, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.65097961425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6153930664062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27140, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2005376324057579, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.11755337789654732, + "grad_norm": 0.03838497772812843, + "kept_token_count": 654.0203125, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925502896308899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074497040361166, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.93443908691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5148071289062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27150, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20074577927589415, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.11687366962432862, + "grad_norm": 0.04443897306919098, + "kept_token_count": 654.0984375, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925670862197876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074328936636447905, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0534973144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9388946533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27160, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19976872503757476, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.12986042946577073, + "grad_norm": 0.041243698447942734, + "kept_token_count": 654.98125, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928738355636597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0071261503733694555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0979522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.22194519042966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27170, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.20009805411100387, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.12398974597454071, + "grad_norm": 0.03696415200829506, + "kept_token_count": 654.4046875, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921646893024445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007835311535745858, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3341522216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.62714233398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27180, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.1995508924126625, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.12092786654829979, + "grad_norm": 0.0530395470559597, + "kept_token_count": 655.090625, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923856496810913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007614338584244251, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0845611572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.94358520507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27190, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20016915053129197, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.13249530494213105, + "grad_norm": 0.05377564951777458, + "kept_token_count": 654.6421875, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.992253702878952, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007746300473809242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7946380615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7586608886719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27200, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.19952951669692992, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.1235746130347252, + "grad_norm": 0.04969455301761627, + "kept_token_count": 655.171875, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922088265419007, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007791167311370373, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.16375732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.34117431640624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27210, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19974044263362883, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.11893921345472336, + "grad_norm": 0.05525915324687958, + "kept_token_count": 654.8234375, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926801085472107, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007319904770702124, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.63832092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.8863250732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19913201332092284, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.12494208738207817, + "grad_norm": 0.046631891280412674, + "kept_token_count": 655.3953125, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924359023571014, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00756410313770175, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4869049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8135498046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27230, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1999882847070694, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.12778556421399118, + "grad_norm": 0.04349489137530327, + "kept_token_count": 654.6109375, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926815390586853, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007318452373147011, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9772705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8877288818359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27240, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19990739822387696, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.1303958147764206, + "grad_norm": 0.03595636039972305, + "kept_token_count": 654.6359375, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928178131580353, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007182194665074348, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4256195068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3402954101563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20070005506277083, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.12945801094174386, + "grad_norm": 0.037338271737098694, + "kept_token_count": 653.9953125, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923225164413452, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007677497807890177, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.55462646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9549591064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27260, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.199652236700058, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.12095039561390877, + "grad_norm": 0.03564569726586342, + "kept_token_count": 654.86875, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927014589309693, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007298543397337198, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4660278320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7352020263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27270, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20052045583724976, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.1290852427482605, + "grad_norm": 0.04221280664205551, + "kept_token_count": 654.278125, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923688352108002, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076311580836772915, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9439270019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1125518798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27280, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20054661631584167, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.12671858295798302, + "grad_norm": 0.08491536229848862, + "kept_token_count": 653.9046875, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925658822059631, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074341089464724066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6428588867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4733093261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27290, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.19904222786426545, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.11919903829693794, + "grad_norm": 0.04979082569479942, + "kept_token_count": 655.3640625, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924851179122924, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007514890842139721, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.50865478515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.16412048339845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27300, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.1991562306880951, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.13398523926734923, + "grad_norm": 0.036187876015901566, + "kept_token_count": 655.375, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926854431629181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007314557861536741, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.69215698242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3472534179688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27310, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19917396157979966, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.13022529408335687, + "grad_norm": 0.04132182523608208, + "kept_token_count": 655.4234375, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928961873054505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007103806640952825, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2332000732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.056396484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27320, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.20099648088216782, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.12924908697605134, + "grad_norm": 0.0472368560731411, + "kept_token_count": 653.6234375, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0068, + "mgd_cos_sim": 0.9931756556034088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006824336387217045, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4582763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4279083251953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27330, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.19917364716529845, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.1310335576534271, + "grad_norm": 0.03303065896034241, + "kept_token_count": 655.25, + "learning_rate": 6.406855296131387e-07, + "loss": 0.008, + "mgd_cos_sim": 0.9919975876808167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008002413995563984, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.45005493164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.95566711425784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27340, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1998738795518875, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.12241743505001068, + "grad_norm": 0.03541398048400879, + "kept_token_count": 654.6796875, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928518891334533, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0071481048129498955, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9432434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9825012207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27350, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19981522411108016, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.11935253888368606, + "grad_norm": 0.04511313512921333, + "kept_token_count": 654.996875, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925770580768585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007422943226993084, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4621246337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.51068115234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27360, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.19949951171875, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.12061761617660523, + "grad_norm": 0.0534839853644371, + "kept_token_count": 655.01875, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925193607807159, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007480645924806595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2882598876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0852294921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27370, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20013774037361146, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.1212875634431839, + "grad_norm": 0.052058227360248566, + "kept_token_count": 654.3625, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9922730624675751, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0077269296161830425, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3937713623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.84247131347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27380, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.19940063059329988, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.12711675241589546, + "grad_norm": 0.047315485775470734, + "kept_token_count": 655.0, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923088133335114, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007691190391778946, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.27411499023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.89222717285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27390, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.20089719146490098, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.12770194560289383, + "grad_norm": 0.0526345893740654, + "kept_token_count": 653.8796875, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927130997180938, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007286899629980326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.356103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.523095703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.19944746941328048, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.1243758112192154, + "grad_norm": 0.049960918724536896, + "kept_token_count": 654.95, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.9920672357082367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00793276485055685, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9683868408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.488037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27410, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19998642057180405, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.12629347443580627, + "grad_norm": 0.04163345322012901, + "kept_token_count": 654.271875, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925419390201569, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00745804226025939, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6571563720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67034606933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27420, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.20047730058431626, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.1311293751001358, + "grad_norm": 0.03771715611219406, + "kept_token_count": 654.1359375, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926543235778809, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007345679588615895, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1943023681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1910736083984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27430, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.19969925433397293, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.12670367434620858, + "grad_norm": 0.040013138204813004, + "kept_token_count": 654.834375, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927460312843323, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007253971043974161, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.73724975585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.18902587890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27440, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19951896667480468, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.13284190520644187, + "grad_norm": 0.047626834362745285, + "kept_token_count": 655.028125, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925799489021301, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007420055754482746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8539672851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7309844970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27450, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20040908008813857, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.12358996272087097, + "grad_norm": 0.04302824288606644, + "kept_token_count": 654.3484375, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925543904304505, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074456027708947655, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5227905273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.1878265380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27460, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19998729825019837, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.12641831412911414, + "grad_norm": 0.05417711287736893, + "kept_token_count": 654.5484375, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925795674324036, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007420423906296492, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.95081787109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6964935302734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27470, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2000693276524544, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.13499677628278733, + "grad_norm": 0.0496557392179966, + "kept_token_count": 654.6828125, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926335394382477, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007366455253213644, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7289245605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.36700134277345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27480, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20017605125904084, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.1241113819181919, + "grad_norm": 0.039124995470047, + "kept_token_count": 654.265625, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0065, + "mgd_cos_sim": 0.9934905230998993, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006509472895413637, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.87523193359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.26189575195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27490, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.19992446452379226, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.119967220723629, + "grad_norm": 0.05050530284643173, + "kept_token_count": 654.7890625, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.992659968137741, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073400335386395454, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.37620239257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4501037597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27500, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19965934604406357, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.12576893121004104, + "grad_norm": 0.055272314697504044, + "kept_token_count": 654.6328125, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.9921261310577393, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007873871363699436, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4960632324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7621124267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27510, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.2002149060368538, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.13079985231161118, + "grad_norm": 0.03643958643078804, + "kept_token_count": 654.540625, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925488710403443, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007451145350933075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.83489379882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1647644042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27520, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19990367591381072, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.12135629802942276, + "grad_norm": 0.04876590147614479, + "kept_token_count": 654.89375, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923140347003937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007685956824570894, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6849090576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.2018096923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27530, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.20027219653129577, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.12508131116628646, + "grad_norm": 0.035953667014837265, + "kept_token_count": 654.6390625, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927118957042694, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007288102991878986, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.38629150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27540, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.19984148740768432, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.11916370019316673, + "grad_norm": 0.045987214893102646, + "kept_token_count": 654.7265625, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924383342266083, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007561660744249821, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3449462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7053253173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27550, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.19990562200546264, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.1289570800960064, + "grad_norm": 0.0507369264960289, + "kept_token_count": 654.65625, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924919128417968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007508079055696726, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8451385498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.63657531738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27560, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1992607519030571, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.12748439237475395, + "grad_norm": 0.050982095301151276, + "kept_token_count": 655.4578125, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928407490253448, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00715923523530364, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.65374450683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3904052734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27570, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.20029374659061433, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.12349683046340942, + "grad_norm": 0.04642510041594505, + "kept_token_count": 654.55625, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924744606018067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007525544986128807, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7435607910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.32558288574216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27580, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20006263852119446, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.1127500556409359, + "grad_norm": 0.0349651537835598, + "kept_token_count": 654.4484375, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925159454345703, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007484046649187804, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2309051513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.77843322753904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27590, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20012354850769043, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.11659222319722176, + "grad_norm": 0.044019341468811035, + "kept_token_count": 654.7203125, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928098738193512, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0071901284158229824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.14185791015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.2354278564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27600, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.20014562457799911, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.12356388345360755, + "grad_norm": 0.04778121039271355, + "kept_token_count": 654.4484375, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927871763706207, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007212824560701847, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4391571044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.95070495605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27610, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19922526627779008, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.12604344561696051, + "grad_norm": 0.033394597470760345, + "kept_token_count": 655.5140625, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926667988300324, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073331986553966996, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6629608154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2167633056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27620, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.20011406838893891, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.13776940032839774, + "grad_norm": 0.047134265303611755, + "kept_token_count": 654.696875, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921914041042328, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007808589655905962, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2445343017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.55254821777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27630, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.20092723667621612, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.1274675652384758, + "grad_norm": 0.035439249128103256, + "kept_token_count": 653.940625, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923926949501037, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007607305329293013, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6234100341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.1764312744141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27640, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.1994667112827301, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.1227408692240715, + "grad_norm": 0.045153241604566574, + "kept_token_count": 655.0359375, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925633728504181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007436629198491573, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9913055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.67937622070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27650, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20015717446804046, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.13044437170028686, + "grad_norm": 0.04707302153110504, + "kept_token_count": 654.6984375, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923307359218597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007669255416840315, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7678924560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1700500488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27660, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.1995043620467186, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.1235455520451069, + "grad_norm": 0.08114704489707947, + "kept_token_count": 654.925, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923104524612427, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007689549122005701, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.08033447265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5524200439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27670, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2000410556793213, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.13171895444393159, + "grad_norm": 0.052662767469882965, + "kept_token_count": 654.4203125, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928193032741547, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007180692069232464, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0257995605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2596466064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27680, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.20020989328622818, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.12551825866103172, + "grad_norm": 0.04021468758583069, + "kept_token_count": 654.571875, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921776473522186, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007822342030704022, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.79308166503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4455169677734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27690, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20008121132850648, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.1198146902024746, + "grad_norm": 0.039375223219394684, + "kept_token_count": 654.2234375, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926240086555481, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007375997770577669, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0247528076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1415161132812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27700, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.20036910772323607, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.12835128903388976, + "grad_norm": 0.036426883190870285, + "kept_token_count": 654.078125, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922473847866058, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007752606738358736, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.70807189941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9293548583984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27710, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.20075669288635253, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.14059156328439712, + "grad_norm": 0.045547809451818466, + "kept_token_count": 654.0328125, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922401368618011, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007759849261492491, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0990386962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59978637695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.20039790570735933, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.13616078943014145, + "grad_norm": 0.052330195903778076, + "kept_token_count": 654.05625, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9922906577587127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007709331251680851, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1706207275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.5919464111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27730, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.2000574752688408, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.12859946936368943, + "grad_norm": 0.039225026965141296, + "kept_token_count": 654.44375, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927783131599426, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007221671286970377, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9829467773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8160858154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2009223833680153, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.12446467280387878, + "grad_norm": 0.05043020844459534, + "kept_token_count": 653.7078125, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.992323100566864, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00767689598724246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.66669006347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.51112670898436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27750, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20014849603176116, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.11683635264635087, + "grad_norm": 0.04952447861433029, + "kept_token_count": 654.6078125, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926144897937774, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007385488227009773, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1249542236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.68993225097654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27760, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.200681671500206, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.1161517933011055, + "grad_norm": 0.05057235434651375, + "kept_token_count": 654.1203125, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.992521983385086, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007478025835007429, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1812225341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6784912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27770, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19975378066301347, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.13407719805836676, + "grad_norm": 0.057652927935123444, + "kept_token_count": 654.8015625, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925931096076965, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007406884618103504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8148468017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4405181884766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27780, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20045674592256546, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.12223332822322845, + "grad_norm": 0.0560959093272686, + "kept_token_count": 654.365625, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924264788627625, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007573530171066522, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3921600341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.35841369628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27790, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.20028092414140702, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.12471865639090537, + "grad_norm": 0.04347333312034607, + "kept_token_count": 654.303125, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927651941776275, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007234799489378929, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0253204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1995330810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27800, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.20059770196676255, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.135374616086483, + "grad_norm": 0.03982989862561226, + "kept_token_count": 654.015625, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923039138317108, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007696102745831013, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8319458007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.31626586914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27810, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19909568578004838, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.1324910894036293, + "grad_norm": 0.05114417523145676, + "kept_token_count": 655.328125, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926829993724823, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073170092888176445, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5765350341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1897674560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27820, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19996431320905686, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.12830577716231345, + "grad_norm": 0.04853937029838562, + "kept_token_count": 654.628125, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0082, + "mgd_cos_sim": 0.9918182253837585, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00818175682798028, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.99363098144534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.87171325683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27830, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20068450570106505, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.11891551613807679, + "grad_norm": 0.05577889084815979, + "kept_token_count": 653.9625, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0083, + "mgd_cos_sim": 0.991687285900116, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008312712050974369, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7976928710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05608520507815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27840, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19997387528419494, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.1282414399087429, + "grad_norm": 0.039943479001522064, + "kept_token_count": 654.709375, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923708140850067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007629206404089928, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.68580932617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6728088378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.20044529736042022, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.11890296190977097, + "grad_norm": 0.0529121570289135, + "kept_token_count": 654.4546875, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923282742500306, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007671722117811442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.78138122558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.16533813476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27860, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.19956160336732864, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.12390152961015702, + "grad_norm": 0.036851078271865845, + "kept_token_count": 654.8953125, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922485589981079, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007751443050801754, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7377227783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7394287109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27870, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20010669380426407, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.12922823205590248, + "grad_norm": 0.04888857156038284, + "kept_token_count": 654.4609375, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926969170570373, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073030821047723295, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2974395751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.00572814941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27880, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20018995255231858, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.1258535534143448, + "grad_norm": 0.04487534239888191, + "kept_token_count": 654.1484375, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922030389308929, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007796969916671515, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.81334533691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3556274414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27890, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.19990803450345992, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.13019343540072442, + "grad_norm": 0.06247812137007713, + "kept_token_count": 654.728125, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924430251121521, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007556973956525326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2921661376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.92400817871095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27900, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20018249601125718, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.12966799288988112, + "grad_norm": 0.037390947341918945, + "kept_token_count": 654.265625, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928222596645355, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007177730184048414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5116241455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3517669677734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27910, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20024122148752213, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.13332556188106537, + "grad_norm": 0.04552542045712471, + "kept_token_count": 654.2375, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.992652851343155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007347163092344999, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5411071777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5766082763672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27920, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20028512477874755, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.12112476453185081, + "grad_norm": 0.05740644037723541, + "kept_token_count": 654.3640625, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921547412872315, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007845251075923442, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7477752685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2297790527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27930, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.1997231125831604, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.11795889139175415, + "grad_norm": 0.04972062632441521, + "kept_token_count": 654.853125, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926238000392914, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00737619362771511, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1222778320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9971160888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20000969022512435, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.12839143872261047, + "grad_norm": 0.04966633394360542, + "kept_token_count": 654.4703125, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928545475006103, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007145457435399294, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.54100036621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.59413757324216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27950, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19965311586856843, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.13497389778494834, + "grad_norm": 0.04964349418878555, + "kept_token_count": 654.771875, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927021741867066, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007297815941274166, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0231872558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.258740234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27960, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.1998076856136322, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.13474742621183394, + "grad_norm": 0.04930669069290161, + "kept_token_count": 654.775, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0083, + "mgd_cos_sim": 0.9916984498500824, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008301554899662732, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6840484619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1714263916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19981686323881148, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.12366138845682144, + "grad_norm": 0.043914128094911575, + "kept_token_count": 654.56875, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927163064479828, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00728369653224945, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0078643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2232940673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27980, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19985247552394866, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.12645996659994124, + "grad_norm": 0.042790330946445465, + "kept_token_count": 654.721875, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.9921460688114166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00785392476245761, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.98924560546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.42681884765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27990, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19968059062957763, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.12036444544792176, + "grad_norm": 0.07410449534654617, + "kept_token_count": 654.9484375, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.99205362200737, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007946368027478456, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.44041137695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.67579040527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28000, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19967710971832275, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.13542468324303628, + "grad_norm": 0.059978824108839035, + "kept_token_count": 654.88125, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926998019218445, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007300180196762085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.58558349609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31329956054685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28010, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19990540891885758, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.1255223870277405, + "grad_norm": 0.04099611937999725, + "kept_token_count": 654.6859375, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928761959075928, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00712380614131689, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6394805908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0400024414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20017399191856383, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.12895984724164009, + "grad_norm": 0.04279573634266853, + "kept_token_count": 654.5265625, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923782587051392, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007621734961867332, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7432373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7255126953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28030, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2003087133169174, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.1305747263133526, + "grad_norm": 0.05119674280285835, + "kept_token_count": 654.30625, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.007, + "mgd_cos_sim": 0.9930325984954834, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006967406533658504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7657745361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.22022705078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28040, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.20041273683309554, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.13138926550745963, + "grad_norm": 0.04804025962948799, + "kept_token_count": 654.340625, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927569329738617, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007243064977228641, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8628692626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8982177734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28050, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20069606602191925, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.1356813944876194, + "grad_norm": 0.043756429105997086, + "kept_token_count": 654.03125, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.992835009098053, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007164993695914746, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.00493774414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5325134277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28060, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2008301645517349, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.11701718419790268, + "grad_norm": 0.053923141211271286, + "kept_token_count": 654.0703125, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925105988979339, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007489403616636992, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.61664428710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.89109497070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28070, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20047618597745895, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.13937490433454514, + "grad_norm": 0.050026483833789825, + "kept_token_count": 654.0359375, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923742413520813, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007625739928334952, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.18402099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3006561279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28080, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.2001881331205368, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.1316543348133564, + "grad_norm": 0.06639450788497925, + "kept_token_count": 654.45, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9929153084754944, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007084668520838023, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.98509216308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94969482421874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28090, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.1997382014989853, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.12386270985007286, + "grad_norm": 0.03648972138762474, + "kept_token_count": 654.8234375, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924391150474549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007560880295932293, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9892181396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0841949462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28100, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20021683126688003, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.1251952603459358, + "grad_norm": 0.05551578477025032, + "kept_token_count": 654.421875, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.008, + "mgd_cos_sim": 0.9920324504375457, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007967548165470361, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7400299072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0590576171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28110, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19996449053287507, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.12364798784255981, + "grad_norm": 0.05224335193634033, + "kept_token_count": 654.5625, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927199482917786, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007280051056295633, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0006866455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28120, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19909347146749495, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.13357440903782844, + "grad_norm": 0.04283035174012184, + "kept_token_count": 655.540625, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9929055392742157, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007094460818916559, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8586730957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7409210205078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28130, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19888088703155518, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.12209843397140503, + "grad_norm": 0.03487623855471611, + "kept_token_count": 655.61875, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927489459514618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007251037936657667, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9978881835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4952423095703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1999468356370926, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.11765066459774971, + "grad_norm": 0.038708560168743134, + "kept_token_count": 654.6671875, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922310292720795, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00776898292824626, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3950164794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.6051055908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28150, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20064045041799544, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.13548186272382737, + "grad_norm": 0.034366674721241, + "kept_token_count": 654.2046875, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924206018447876, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007579394895583391, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6453155517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1638946533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28160, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20014267712831496, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.1264777585864067, + "grad_norm": 0.04332119598984718, + "kept_token_count": 654.453125, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921919643878937, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0078080427832901474, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.79140319824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.41547241210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28170, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19957067370414733, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.12944265976548194, + "grad_norm": 0.0632973462343216, + "kept_token_count": 655.0875, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927366495132446, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007263372279703617, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7395538330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.006494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20039280652999877, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.1296697199344635, + "grad_norm": 0.043638382107019424, + "kept_token_count": 654.05625, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0069, + "mgd_cos_sim": 0.993071311712265, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006928691826760769, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.85109252929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.58494873046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28190, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.1994042292237282, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.12486119642853737, + "grad_norm": 0.04477835074067116, + "kept_token_count": 655.021875, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928616583347321, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007138331048190593, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.62506713867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1123474121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19902055263519286, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.11886904910206794, + "grad_norm": 0.04009467363357544, + "kept_token_count": 655.478125, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923751354217529, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007624865882098675, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.16331481933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.46551513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28210, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2006609931588173, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.11386110186576844, + "grad_norm": 0.04370126500725746, + "kept_token_count": 654.5015625, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928005456924438, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007199449371546507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6924163818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.38006286621095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28220, + "valid_token_count": 818.796875 + }, + { + "actual_token_mask_ratio": 0.19971197247505187, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.11953356340527535, + "grad_norm": 0.0399908572435379, + "kept_token_count": 654.890625, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926726937294006, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007327313348650932, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.18336181640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67039794921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28230, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.20084306895732879, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.12576666846871376, + "grad_norm": 0.04180291295051575, + "kept_token_count": 654.025, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.992582643032074, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074173700995743275, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1768402099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67493591308596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28240, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19952934980392456, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.12719869762659072, + "grad_norm": 0.05613723397254944, + "kept_token_count": 654.9078125, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925683796405792, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007431639730930329, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0692413330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5851287841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28250, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20007958263158798, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.12440143153071404, + "grad_norm": 0.037379875779151917, + "kept_token_count": 654.453125, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.992835932970047, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007164077647030353, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3389099121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5808929443359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28260, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.20015270709991456, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.12686299309134483, + "grad_norm": 0.04371203854680061, + "kept_token_count": 654.5578125, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925680160522461, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007431985717266798, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3398773193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5433837890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28270, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20036090016365052, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.1257658541202545, + "grad_norm": 0.037104565650224686, + "kept_token_count": 654.0640625, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923924088478089, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007607586402446032, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.68840942382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.48170776367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28280, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.1988326132297516, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.12145282253623009, + "grad_norm": 0.038737110793590546, + "kept_token_count": 655.5640625, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.992628812789917, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007371181808412075, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.3015655517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.4024230957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28290, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19903053641319274, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.12137913629412651, + "grad_norm": 0.04043438658118248, + "kept_token_count": 655.33125, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921689510345459, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00783105343580246, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0881683349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.59735107421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28300, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20055516362190245, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.13803680539131163, + "grad_norm": 0.046784959733486176, + "kept_token_count": 654.290625, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924972891807556, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007502704486250877, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7197204589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7446594238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28310, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.1999144196510315, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.11669388338923455, + "grad_norm": 0.04321253299713135, + "kept_token_count": 654.359375, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925594449043273, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007440565619617701, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.54502563476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9081329345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28320, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.19954150170087814, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.12452879026532174, + "grad_norm": 0.048922110348939896, + "kept_token_count": 654.9046875, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.992291533946991, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007708467822521925, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0004638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.18321533203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28330, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2003890410065651, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.12572221606969833, + "grad_norm": 0.05159792676568031, + "kept_token_count": 654.3671875, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.008, + "mgd_cos_sim": 0.9919561505317688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008043863344937563, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0491882324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9508056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28340, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2001255437731743, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.11322317272424698, + "grad_norm": 0.03219893202185631, + "kept_token_count": 654.728125, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.99253631234169, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007463680114597082, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0776336669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.37286071777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28350, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.19955408275127412, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.12789329439401625, + "grad_norm": 0.0467936210334301, + "kept_token_count": 654.7921875, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923042356967926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00769577082246542, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5432098388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4966156005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28360, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20108328610658646, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.12917228266596795, + "grad_norm": 0.05339142307639122, + "kept_token_count": 653.5125, + "learning_rate": 2.41773707498531e-07, + "loss": 0.008, + "mgd_cos_sim": 0.9919544816017151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008045510668307543, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4370391845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5995330810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28370, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.19962908327579498, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.1277470774948597, + "grad_norm": 0.06513708084821701, + "kept_token_count": 654.709375, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924033164978028, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007596687134355307, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4520660400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.69142761230466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28380, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.19976394325494767, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.12555609792470931, + "grad_norm": 0.042787350714206696, + "kept_token_count": 654.803125, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0067, + "mgd_cos_sim": 0.9933498978614808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006650097481906414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5872344970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7754638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28390, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19957518577575684, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.11689341813325882, + "grad_norm": 0.06169901788234711, + "kept_token_count": 655.075, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922425746917725, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007757426239550114, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2559326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.17348022460936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28400, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.1991952806711197, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.12574202343821525, + "grad_norm": 0.03574233874678612, + "kept_token_count": 655.359375, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.992577713727951, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007422294095158577, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7071258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.80438232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28410, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.1998676836490631, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.1228800430893898, + "grad_norm": 0.0427766889333725, + "kept_token_count": 654.946875, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9928900897502899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007109907548874616, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1200866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.36676025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28420, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.19980357885360717, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.12808116003870965, + "grad_norm": 0.053811028599739075, + "kept_token_count": 654.840625, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926308512687683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007369143236428499, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0638061523438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8379852294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.19934274405241012, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.12324427664279938, + "grad_norm": 0.07914141565561295, + "kept_token_count": 655.3, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923937082290649, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007606292515993119, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1343292236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1051422119141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28440, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.19962792247533798, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.12464421391487121, + "grad_norm": 0.048860713839530945, + "kept_token_count": 655.06875, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927287757396698, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007271210104227066, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.46140747070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8526306152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.2003689393401146, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.11727184131741523, + "grad_norm": 0.05198968946933746, + "kept_token_count": 654.5171875, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925334036350251, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0074665994383394715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.95460815429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.21844482421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28460, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.20128130912780762, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.13303087577223777, + "grad_norm": 0.049294669181108475, + "kept_token_count": 653.3828125, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925063192844391, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007493677549064159, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6688934326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1261901855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28470, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.200177401304245, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.12830328494310378, + "grad_norm": 0.04520128294825554, + "kept_token_count": 654.534375, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924835681915283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007516432367265224, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.67774353027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1197448730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19953305423259735, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.1246152326464653, + "grad_norm": 0.05516998469829559, + "kept_token_count": 654.9140625, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0079, + "mgd_cos_sim": 0.992067015171051, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00793299013748765, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6385559082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4995391845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28490, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1995500147342682, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.12600201666355132, + "grad_norm": 0.037485506385564804, + "kept_token_count": 655.0984375, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0082, + "mgd_cos_sim": 0.9918080985546112, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008191895950585604, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0046020507813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28500, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.19975612163543702, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.1205398254096508, + "grad_norm": 0.04363642632961273, + "kept_token_count": 654.6734375, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923800647258758, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007619939744472504, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1190216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5798858642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28510, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.20016174465417863, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.12426899299025536, + "grad_norm": 0.04030674323439598, + "kept_token_count": 654.675, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925205409526825, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007479451224207878, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0755310058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9042907714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28520, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.1991964042186737, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.1166787639260292, + "grad_norm": 0.038988638669252396, + "kept_token_count": 655.2078125, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0069, + "mgd_cos_sim": 0.9931041181087494, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006895877420902252, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4019500732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1994110107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28530, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.20044871270656586, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.13022931516170502, + "grad_norm": 0.03919095918536186, + "kept_token_count": 654.3390625, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.007, + "mgd_cos_sim": 0.9930326521396637, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006967351865023374, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.92810974121096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28540, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.20018049478530883, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.11907551735639572, + "grad_norm": 0.042560894042253494, + "kept_token_count": 654.3375, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9922652184963227, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007734781224280595, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.69615478515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8317474365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28550, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.1996777579188347, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.11863590255379677, + "grad_norm": 0.044694263488054276, + "kept_token_count": 655.0515625, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.007, + "mgd_cos_sim": 0.993044912815094, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0069550852291285995, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6294860839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.74778747558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28560, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20101915746927262, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.12699032500386237, + "grad_norm": 0.059984948486089706, + "kept_token_count": 653.878125, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925776124000549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007422383595257998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5472717285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6577453613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28570, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19981628209352492, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.1301168031990528, + "grad_norm": 0.04841664806008339, + "kept_token_count": 654.7359375, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925531864166259, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007446823269128799, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0229949951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.29755249023435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28580, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.1999830961227417, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.1250613287091255, + "grad_norm": 0.050168268382549286, + "kept_token_count": 654.6234375, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925539553165436, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007446031924337149, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5552032470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2740112304688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28590, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19950556606054307, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.12459812387824058, + "grad_norm": 0.03768946975469589, + "kept_token_count": 655.2640625, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924775660037994, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007522452622652054, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0825653076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.76153564453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28600, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.19960466772317886, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.12772431448101998, + "grad_norm": 0.059450604021549225, + "kept_token_count": 655.0921875, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9927821278572082, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007217892445623875, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9357971191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.278857421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19963269084692, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.1254081755876541, + "grad_norm": 0.05331715941429138, + "kept_token_count": 654.9609375, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924070000648498, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007593007199466229, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0152618408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0336486816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28620, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2000840812921524, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.1298096865415573, + "grad_norm": 0.037314970046281815, + "kept_token_count": 654.6296875, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0071, + "mgd_cos_sim": 0.9929355084896088, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007064495421946049, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1860412597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81700439453124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28630, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.20014095604419707, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.1267055779695511, + "grad_norm": 0.05280754342675209, + "kept_token_count": 654.5484375, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.99219930768013, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007800703216344118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1474670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.60975952148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28640, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20005767792463303, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.12495330050587654, + "grad_norm": 0.06138747185468674, + "kept_token_count": 654.4046875, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926247656345367, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007375223468989134, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.67523193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49241333007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28650, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2005129337310791, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.13019177317619324, + "grad_norm": 0.049117039889097214, + "kept_token_count": 653.8921875, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923164546489716, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007683555874973535, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9844940185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.71363220214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28660, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.20036156773567199, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.12751275002956391, + "grad_norm": 0.03601633012294769, + "kept_token_count": 654.1015625, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0069, + "mgd_cos_sim": 0.993080872297287, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006919144093990326, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7477691650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.46531066894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28670, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20050169229507447, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.12150724828243256, + "grad_norm": 0.06492911279201508, + "kept_token_count": 654.35625, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.99282266497612, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00717734694480896, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7869049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.13409423828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28680, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.2003531873226166, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.12677888721227645, + "grad_norm": 0.05114532262086868, + "kept_token_count": 654.3625, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9923759996891022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007624011393636465, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7508056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3500183105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28690, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2010040283203125, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.12503628730773925, + "grad_norm": 0.04308585450053215, + "kept_token_count": 653.946875, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927277982234954, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007272197678685188, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8214874267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4023468017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28700, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.19966395497322081, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.12730256021022796, + "grad_norm": 0.03776547685265541, + "kept_token_count": 654.8, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.99262934923172, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007370656728744507, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4728607177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.60996398925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28710, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19895787090063094, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.1287457153201103, + "grad_norm": 0.03397443890571594, + "kept_token_count": 655.4796875, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9922284781932831, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007771521806716919, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.43506774902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.49510803222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28720, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2000734254717827, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.12301702797412872, + "grad_norm": 0.035722434520721436, + "kept_token_count": 654.8859375, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.992758595943451, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007241416536271572, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7873931884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.11472473144534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28730, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.20090753883123397, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.12543298080563545, + "grad_norm": 0.048224013298749924, + "kept_token_count": 653.8671875, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928495347499847, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0071504688821733, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9270782470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.330859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20059768855571747, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.13339065611362458, + "grad_norm": 0.05078212916851044, + "kept_token_count": 654.0, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.992746114730835, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007253882009536028, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2587493896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5372741699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28750, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19980833530426026, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.1289967678487301, + "grad_norm": 0.03836173564195633, + "kept_token_count": 654.6859375, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0076, + "mgd_cos_sim": 0.9924445509910583, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007555449195206165, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2334899902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9319152832031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28760, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.19991112053394317, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.12791696563363075, + "grad_norm": 0.05147867277264595, + "kept_token_count": 654.9375, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924921810626983, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007507814653217792, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.65114135742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1440948486328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28770, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.19945162683725357, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.12509193345904351, + "grad_norm": 0.04320944473147392, + "kept_token_count": 655.1359375, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.992226105928421, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007773889414966107, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7659545898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5706848144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28780, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.200117826461792, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.12838749438524247, + "grad_norm": 0.05472879111766815, + "kept_token_count": 654.5078125, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9926795542240143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007320452388375998, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6056701660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.50628051757815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28790, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20002153664827346, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.13047220557928085, + "grad_norm": 0.03643061965703964, + "kept_token_count": 654.4890625, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9922878444194794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007712163403630257, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0911163330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9884857177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.1995753049850464, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.12871127352118492, + "grad_norm": 0.045284390449523926, + "kept_token_count": 654.890625, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0073, + "mgd_cos_sim": 0.9927469313144683, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007253057416528463, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6332092285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3363372802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28810, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.20061190128326417, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.12018969506025315, + "grad_norm": 0.049613043665885925, + "kept_token_count": 654.203125, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926324784755707, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007367517612874508, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.69691772460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.25123596191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.20128223747015, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.12984250858426094, + "grad_norm": 0.05338359996676445, + "kept_token_count": 653.5875, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924766302108765, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0075233721174299715, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5270233154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8379791259766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20181049555540084, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.12270328849554062, + "grad_norm": 0.04660838097333908, + "kept_token_count": 653.409375, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924819827079773, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007518021762371064, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2865509033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.06483764648436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28840, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.1999562367796898, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.1278555765748024, + "grad_norm": 0.03710074722766876, + "kept_token_count": 654.5015625, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926283836364747, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007371607143431902, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7346435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1801330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28850, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.19979318231344223, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.12163884043693543, + "grad_norm": 0.04676518961787224, + "kept_token_count": 654.815625, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.992784035205841, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007215964049100876, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.219384765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4399169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28860, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20122834742069245, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.12967583909630775, + "grad_norm": 0.037073902785778046, + "kept_token_count": 653.95, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0078, + "mgd_cos_sim": 0.9921771824359894, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007822816073894501, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1003723144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28870, + "valid_token_count": 818.6953125 + }, + { + "actual_token_mask_ratio": 0.20056798607110976, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.128645870834589, + "grad_norm": 0.053235821425914764, + "kept_token_count": 654.35625, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0077, + "mgd_cos_sim": 0.9923396527767181, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007660344522446394, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2519897460937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.04522399902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28880, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.19945183843374253, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.1242351658642292, + "grad_norm": 0.0402272492647171, + "kept_token_count": 654.95, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9925256431102752, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007474350091069937, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.55051574707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3405334472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28890, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20138121843338014, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.13527927100658416, + "grad_norm": 0.07614760100841522, + "kept_token_count": 653.421875, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9926046907901764, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073953231796622275, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.35646057128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7077362060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28900, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20038032829761504, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.12452981695532798, + "grad_norm": 0.05389586463570595, + "kept_token_count": 654.40625, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925988495349884, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007401153072714805, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6871765136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.96042785644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.199822361767292, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.1266329437494278, + "grad_norm": 0.0539192296564579, + "kept_token_count": 654.8125, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.007, + "mgd_cos_sim": 0.992958277463913, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007041715737432242, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2298583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.27442321777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28920, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.20116280466318132, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.13311936482787132, + "grad_norm": 0.04720020666718483, + "kept_token_count": 653.6171875, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0074, + "mgd_cos_sim": 0.9925679922103882, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007432007230818272, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3345153808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.61720581054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28930, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20042850375175475, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.12381654381752014, + "grad_norm": 0.035863105207681656, + "kept_token_count": 654.2921875, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0072, + "mgd_cos_sim": 0.9928228259086609, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007177173160016537, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.24452209472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4027038574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28940, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.20002930611371994, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.13185246139764786, + "grad_norm": 0.042432114481925964, + "kept_token_count": 654.6140625, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0075, + "mgd_cos_sim": 0.9924520432949067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007547952886670828, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9372833251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59417724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28950, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20059848725795745, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.1306421935558319, + "grad_norm": 0.04315482825040817, + "kept_token_count": 653.8875, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9925597190856934, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007440278120338917, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0858489990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.950048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28960, + "valid_token_count": 817.971875 + }, + { + "actual_token_mask_ratio": 0.19891231656074523, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.12606863752007486, + "grad_norm": 0.04541042447090149, + "kept_token_count": 655.575, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0076, + "mgd_cos_sim": 0.9923607885837555, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00763920545578003, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.198291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0247039794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28970, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20028682202100753, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.13347438722848892, + "grad_norm": 0.06696022301912308, + "kept_token_count": 654.2046875, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9925150454044342, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007484952453523874, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6295104980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6357940673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28980, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.19945360720157623, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.11829096525907516, + "grad_norm": 0.033305924385786057, + "kept_token_count": 654.7640625, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0081, + "mgd_cos_sim": 0.9918814361095428, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008118574973195791, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.70874938964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.42389221191405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28990, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.19890767186880112, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.12961898520588874, + "grad_norm": 0.034994740039110184, + "kept_token_count": 655.2921875, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0072, + "mgd_cos_sim": 0.992840051651001, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007159949094057083, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4354278564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2125671386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29000, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.20076059997081758, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.12308874055743217, + "grad_norm": 0.0555768720805645, + "kept_token_count": 653.7515625, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.9922276794910431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007772310357540846, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.56493225097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0894012451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29010, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19997788667678834, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.1322234258055687, + "grad_norm": 0.19224800169467926, + "kept_token_count": 654.8546875, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.9922946274280549, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007705383002758026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.97205505371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.43447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29020, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.20027923583984375, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.12332374081015587, + "grad_norm": 0.04065012186765671, + "kept_token_count": 654.0046875, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.9923485279083252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007651473674923181, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.50592041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.76856994628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29030, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.2011427566409111, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.12396221309900284, + "grad_norm": 0.05634716525673866, + "kept_token_count": 653.5234375, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9926781952381134, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073218210600316525, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.43006591796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4132415771484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29040, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20047007352113724, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.13230012357234955, + "grad_norm": 0.04467611759901047, + "kept_token_count": 654.434375, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.992336118221283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007663869112730026, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1027557373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.58754577636716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29050, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.1995725780725479, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.1257650263607502, + "grad_norm": 0.06388039141893387, + "kept_token_count": 654.728125, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.992237776517868, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007762225531041622, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5276153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.63583984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29060, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.2000885009765625, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.13442905098199845, + "grad_norm": 0.05384259298443794, + "kept_token_count": 654.4875, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9921437084674836, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007856287620961666, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6830139160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.49876098632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29070, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.20081385970115662, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.1297540545463562, + "grad_norm": 0.04622270539402962, + "kept_token_count": 654.146875, + "learning_rate": 7.723559506062849e-08, + "loss": 0.008, + "mgd_cos_sim": 0.9920499324798584, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007950071431696414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.97044982910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1023315429687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29080, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19937196373939514, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.12503110915422438, + "grad_norm": 0.04831152781844139, + "kept_token_count": 655.071875, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0071, + "mgd_cos_sim": 0.9928893566131591, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007110640686005354, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.81549072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4604522705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29090, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2000233680009842, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.12904525101184844, + "grad_norm": 0.04374130815267563, + "kept_token_count": 654.3296875, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.992576390504837, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007423605304211378, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1276153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.70457458496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29100, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.200737464427948, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.13050508201122285, + "grad_norm": 0.051809683442115784, + "kept_token_count": 654.2578125, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9925008535385131, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007499152049422264, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.35859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9315155029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29110, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.19984140396118164, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.13014449775218964, + "grad_norm": 0.0437702052295208, + "kept_token_count": 654.653125, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9921294152736664, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007870579976588488, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7431213378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5042419433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29120, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.19990776628255844, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.1180585004389286, + "grad_norm": 0.0458248145878315, + "kept_token_count": 654.6625, + "learning_rate": 6.908344002193256e-08, + "loss": 0.008, + "mgd_cos_sim": 0.9920006215572357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00799938291311264, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.34120788574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.606591796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29130, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.20020855814218522, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.12507388591766358, + "grad_norm": 0.03469342738389969, + "kept_token_count": 654.4359375, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926139116287231, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00738608343526721, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.23700256347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2043884277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29140, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2006880134344101, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.12252950072288513, + "grad_norm": 0.048973873257637024, + "kept_token_count": 654.1140625, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.9921725511550903, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007827467285096645, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0628967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.98883056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29150, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.19943017214536668, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.12234625294804573, + "grad_norm": 0.049483522772789, + "kept_token_count": 654.9, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0071, + "mgd_cos_sim": 0.9929063022136688, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007093701791018247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6467712402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6075897216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29160, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20008393377065659, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.11853454113006592, + "grad_norm": 0.045170318335294724, + "kept_token_count": 654.5140625, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9925908267498016, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007409178745001554, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.06964721679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.386181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29170, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20091413855552673, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.13377198129892348, + "grad_norm": 0.046110864728689194, + "kept_token_count": 653.903125, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9925545215606689, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007445473968982696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2656280517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1499908447266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29180, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.20020352751016618, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.13073052167892457, + "grad_norm": 0.041971176862716675, + "kept_token_count": 654.55, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.99222332239151, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007776677794754505, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.0029724121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5846893310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29190, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.1992520123720169, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.1161288246512413, + "grad_norm": 0.038618456572294235, + "kept_token_count": 655.1171875, + "learning_rate": 5.843245776785466e-08, + "loss": 0.007, + "mgd_cos_sim": 0.9929921925067902, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007007807306945324, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.55157165527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7584259033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29200, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19995105713605882, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.12054959982633591, + "grad_norm": 0.06543408334255219, + "kept_token_count": 654.8125, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.007, + "mgd_cos_sim": 0.9930158615112304, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006984141282737255, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3203094482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8277862548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29210, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.19959817826747894, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.1285608045756817, + "grad_norm": 0.04130670055747032, + "kept_token_count": 654.971875, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0071, + "mgd_cos_sim": 0.9928967773914337, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007103218138217926, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5839569091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.30470886230466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29220, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.19981570392847062, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.13259500190615653, + "grad_norm": 0.05452023446559906, + "kept_token_count": 654.7171875, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.992564183473587, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007435822021216154, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.86192932128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.94310607910154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20136510580778122, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.13108459562063218, + "grad_norm": 0.056881316006183624, + "kept_token_count": 654.021875, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9924526989459992, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007547299470752477, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9939422607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.873046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29240, + "valid_token_count": 818.9203125 + }, + { + "actual_token_mask_ratio": 0.19898445904254913, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.11886190101504326, + "grad_norm": 0.039633482694625854, + "kept_token_count": 655.446875, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9925909042358398, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007409087941050529, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.45383605957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0107940673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29250, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20081577003002166, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.1310458905994892, + "grad_norm": 0.057873595505952835, + "kept_token_count": 653.9875, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0072, + "mgd_cos_sim": 0.9927865207195282, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00721347713842988, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.0683898925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1353363037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29260, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2004880890250206, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.12064790204167367, + "grad_norm": 0.04507468640804291, + "kept_token_count": 654.1703125, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9927297353744506, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007270262762904167, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5990356445312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29270, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20059742480516435, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.1256475955247879, + "grad_norm": 0.05847017467021942, + "kept_token_count": 654.18125, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9924546480178833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00754536110907793, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.66319274902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.78138732910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29280, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19872398525476456, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.13234504759311677, + "grad_norm": 0.04568718001246452, + "kept_token_count": 655.3953125, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0076, + "mgd_cos_sim": 0.9923657417297364, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007634243741631508, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8982269287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7397033691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29290, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.20043667703866958, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.12004185169935226, + "grad_norm": 0.04982047155499458, + "kept_token_count": 654.1359375, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.9922384262084961, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007761579938232899, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.79356384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0107879638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29300, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19925298243761064, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.1306445650756359, + "grad_norm": 0.06122247502207756, + "kept_token_count": 655.3578125, + "learning_rate": 4.349280298500291e-08, + "loss": 0.008, + "mgd_cos_sim": 0.9920426070690155, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007957395631819963, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.45818786621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.018115234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29310, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19934284836053848, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.13219395726919175, + "grad_norm": 0.059905264526605606, + "kept_token_count": 655.134375, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926129102706909, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007387089636176824, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0552978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4252075195312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29320, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.199754236638546, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.11787009760737419, + "grad_norm": 0.04749208688735962, + "kept_token_count": 654.79375, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.9923009216785431, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00769908856600523, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.99488220214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6552398681641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29330, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20023403614759444, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.12337605506181717, + "grad_norm": 0.03903261199593544, + "kept_token_count": 654.6125, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0072, + "mgd_cos_sim": 0.992755788564682, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007244214415550232, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.33164672851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5723388671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19973980784416198, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.13613165095448493, + "grad_norm": 0.047291792929172516, + "kept_token_count": 654.8765625, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9926860630512238, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007313941605389118, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.02756652832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.86928100585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20020334124565126, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.12958132326602936, + "grad_norm": 0.039436642080545425, + "kept_token_count": 654.3609375, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.992338627576828, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007661376427859068, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8115295410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2658203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29360, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20045493096113204, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.12712266817688941, + "grad_norm": 0.047612033784389496, + "kept_token_count": 654.5484375, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0076, + "mgd_cos_sim": 0.9923643827438354, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076356098987162115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8075866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.989013671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29370, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.19979151487350463, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.12362248376011849, + "grad_norm": 0.04790415242314339, + "kept_token_count": 654.9109375, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9925082862377167, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007491708826273679, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.1603210449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1169158935547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29380, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.19966139048337936, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.12606101632118225, + "grad_norm": 0.04621933400630951, + "kept_token_count": 654.74375, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9925321936607361, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007467794418334961, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.12119750976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.09685974121095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29390, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.20014249682426452, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.12146114930510521, + "grad_norm": 0.04661836847662926, + "kept_token_count": 654.4484375, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9926528692245483, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007347130216658115, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0533050537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.45453186035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29400, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19942841231822966, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.12012677863240243, + "grad_norm": 0.03615310788154602, + "kept_token_count": 655.1234375, + "learning_rate": 3.18194483608103e-08, + "loss": 0.007, + "mgd_cos_sim": 0.9929895520210266, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007010434381663799, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.66539306640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.47222290039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29410, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.2006571799516678, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.13236389607191085, + "grad_norm": 0.04301755130290985, + "kept_token_count": 653.9515625, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.992606395483017, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00739359874278307, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2180450439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.5978240966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29420, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.20016214847564698, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.1370278887450695, + "grad_norm": 0.03855850175023079, + "kept_token_count": 654.7703125, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9921012580394745, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007898739725351333, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5730377197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.0143981933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29430, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.1997780442237854, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.12022187784314156, + "grad_norm": 0.041201863437891006, + "kept_token_count": 654.6171875, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9926714897155762, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007328518945723772, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7566345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8116943359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29440, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2002190589904785, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.12490253522992134, + "grad_norm": 0.03762014955282211, + "kept_token_count": 654.45, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9927156329154968, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007284355536103249, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3664886474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2207427978516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29450, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19968090802431107, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.137758020311594, + "grad_norm": 0.09025286138057709, + "kept_token_count": 654.9625, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.992106419801712, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007893579266965389, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6277099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.92743530273435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29460, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19956152737140656, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.1362495243549347, + "grad_norm": 0.06434634327888489, + "kept_token_count": 655.00625, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926185965538025, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007381381560117006, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9565460205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.580908203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29470, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.20007631182670593, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.12826820239424705, + "grad_norm": 0.04559563472867012, + "kept_token_count": 654.45625, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.9922955572605133, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007704443205147982, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.11732177734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.82793884277345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29480, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20089084208011626, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.12672803848981856, + "grad_norm": 0.04756760597229004, + "kept_token_count": 653.9328125, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9924715399742127, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007528450153768062, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6465637207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7604125976562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29490, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.19992137849330902, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.12538730949163437, + "grad_norm": 0.04901828244328499, + "kept_token_count": 654.5140625, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0081, + "mgd_cos_sim": 0.9918648719787597, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008135134540498256, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9143035888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.59063415527345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29500, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.20086235851049422, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.1258116565644741, + "grad_norm": 0.043225619941949844, + "kept_token_count": 654.0890625, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0072, + "mgd_cos_sim": 0.9927821516990661, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007217852771282196, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1751739501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9323364257813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29510, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.19920568019151688, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.1310830794274807, + "grad_norm": 0.05765141546726227, + "kept_token_count": 654.959375, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0077, + "mgd_cos_sim": 0.9923433780670166, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0076566247269511225, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.69020080566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0321899414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29520, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.19951936453580857, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.13077396303415298, + "grad_norm": 0.045042239129543304, + "kept_token_count": 654.8671875, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926194727420807, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007380522694438696, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5106689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.01544799804685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29530, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.19986913502216339, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.13169458732008935, + "grad_norm": 0.0645083636045456, + "kept_token_count": 654.659375, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.992496520280838, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007503490056842565, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7588409423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.03387756347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29540, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19956036806106567, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.12346380352973937, + "grad_norm": 0.04209354892373085, + "kept_token_count": 655.0625, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9921126902103424, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007887300197035074, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.7668792724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3276702880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29550, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.20081102401018142, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.12916471362113952, + "grad_norm": 0.05824824050068855, + "kept_token_count": 653.9859375, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.992477399110794, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.00752261308953166, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.02464904785154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2855682373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29560, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19994561672210692, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.1269811362028122, + "grad_norm": 0.049771662801504135, + "kept_token_count": 654.6046875, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926154792308808, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007384527288377285, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.11376037597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9728546142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29570, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20064388811588288, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.13077669367194175, + "grad_norm": 0.04530494660139084, + "kept_token_count": 654.2015625, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926064550876618, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073935437016189095, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.725927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8146942138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29580, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2002414956688881, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.12770442441105842, + "grad_norm": 0.05135859176516533, + "kept_token_count": 654.5015625, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0075, + "mgd_cos_sim": 0.9924812555313111, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007518730405718088, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.1875701904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5364166259766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29590, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2010848343372345, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.13065457344055176, + "grad_norm": 0.045550305396318436, + "kept_token_count": 653.5828125, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0078, + "mgd_cos_sim": 0.9921979308128357, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007802082132548094, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9269287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5291046142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29600, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19970083832740784, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.12446927651762962, + "grad_norm": 0.07906007766723633, + "kept_token_count": 654.96875, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9921447813510895, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007855205051600932, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.83547058105466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.74058532714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29610, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20015189945697784, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.13331465870141984, + "grad_norm": 0.04251618683338165, + "kept_token_count": 654.4953125, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0079, + "mgd_cos_sim": 0.9920751929283143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007924811076372862, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1348541259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8769958496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29620, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20019426345825195, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.12515978142619133, + "grad_norm": 0.042110417038202286, + "kept_token_count": 654.403125, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0074, + "mgd_cos_sim": 0.9926101624965668, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007389842066913843, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1731414794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7969085693359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29630, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20004942566156386, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.1286969907581806, + "grad_norm": 0.07115519046783447, + "kept_token_count": 654.4703125, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0073, + "mgd_cos_sim": 0.9926722466945648, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007327759545296431, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6273651123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05548095703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29640, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20032580196857452, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.12369575276970864, + "grad_norm": 0.04330690577626228, + "kept_token_count": 654.4671875, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0076, + "mgd_cos_sim": 0.9924094080924988, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007590588182210922, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1899047851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.4755889892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29650, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19972918182611465, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.12656865268945694, + "grad_norm": 0.04725023731589317, + "kept_token_count": 654.9609375, + "learning_rate": 1.059568187914861e-08, + "loss": 0.007, + "mgd_cos_sim": 0.9929547309875488, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007045293226838112, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.92815856933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.16527404785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29660, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.19940030425786973, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.1187044233083725, + "grad_norm": 0.05194149911403656, + "kept_token_count": 655.2140625, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9926933825016022, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007306629698723555, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4546783447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.64586486816404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29670, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20050828009843827, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.12107772827148437, + "grad_norm": 0.04531778767704964, + "kept_token_count": 654.209375, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0076, + "mgd_cos_sim": 0.9924467265605926, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007553271669894457, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.56763610839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0726684570312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29680, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20032418221235276, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.1271367609500885, + "grad_norm": 0.04641448333859444, + "kept_token_count": 654.340625, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9926795542240143, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007320462819188833, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2526916503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.64194641113284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2004983752965927, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.12196327745914459, + "grad_norm": 0.05848746374249458, + "kept_token_count": 654.303125, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0077, + "mgd_cos_sim": 0.9923338890075684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007666108850389719, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9082763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2760986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29700, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20021556168794633, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.12416893392801284, + "grad_norm": 0.043193235993385315, + "kept_token_count": 654.4046875, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0077, + "mgd_cos_sim": 0.9922577381134033, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007742254622280597, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.7856781005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.57811584472654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20066614896059037, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.11878740862011909, + "grad_norm": 0.03970080614089966, + "kept_token_count": 654.0921875, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0071, + "mgd_cos_sim": 0.9928676903247833, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007132309954613447, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0170501708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2730987548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29720, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20091512799263, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.12651018798351288, + "grad_norm": 0.05201039835810661, + "kept_token_count": 653.8984375, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9927370607852936, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007262945547699929, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.83019104003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01764526367185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.19933410584926606, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.12116999626159668, + "grad_norm": 0.05444053187966347, + "kept_token_count": 655.375, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9927202761173248, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007279722113162279, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1846069335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.26447143554685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29740, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.1993076518177986, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.1233901970088482, + "grad_norm": 0.04786577448248863, + "kept_token_count": 655.396875, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0079, + "mgd_cos_sim": 0.9920688152313233, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007931183930486441, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1740020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.10875244140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29750, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19972671270370485, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.12615759223699569, + "grad_norm": 0.036661598831415176, + "kept_token_count": 654.9265625, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9927230775356293, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007276916690170765, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.75921020507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.73920288085935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29760, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2002389371395111, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.1292744405567646, + "grad_norm": 0.04291090369224548, + "kept_token_count": 654.4015625, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0077, + "mgd_cos_sim": 0.9922697186470032, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007730279304087162, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.28692016601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.65643615722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20050660967826844, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.12659417539834977, + "grad_norm": 0.04441232606768608, + "kept_token_count": 654.2265625, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0077, + "mgd_cos_sim": 0.9923036217689514, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007696371991187334, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1205718994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.05059814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29780, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20001866221427916, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.11918647661805153, + "grad_norm": 0.05551156774163246, + "kept_token_count": 654.5171875, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9927448511123658, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007255142834037542, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5625457763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3965545654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29790, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20032320469617843, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.13078584298491477, + "grad_norm": 0.03851868584752083, + "kept_token_count": 654.121875, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0074, + "mgd_cos_sim": 0.9926130890846252, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0073869101703166965, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1408996582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.31694946289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29800, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.20079001784324646, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.11506310924887657, + "grad_norm": 0.037035826593637466, + "kept_token_count": 653.9484375, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0074, + "mgd_cos_sim": 0.9925864279270172, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007413593120872974, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.59691467285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.3898162841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29810, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.19988947212696076, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.1207675538957119, + "grad_norm": 0.05984675884246826, + "kept_token_count": 654.5625, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0075, + "mgd_cos_sim": 0.9924629330635071, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007537062931805849, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.8387420654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0186706542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29820, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.20084307193756104, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.12803199216723443, + "grad_norm": 0.045036301016807556, + "kept_token_count": 653.9953125, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0071, + "mgd_cos_sim": 0.9928966641426087, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.0071033302694559096, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.70375366210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9767791748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29830, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.1991403356194496, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.12647130861878395, + "grad_norm": 0.03944240137934685, + "kept_token_count": 655.2421875, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0074, + "mgd_cos_sim": 0.9925636053085327, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007436391804367304, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4174346923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9498809814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29840, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1996942490339279, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.13410086035728455, + "grad_norm": 0.03859779238700867, + "kept_token_count": 654.76875, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0076, + "mgd_cos_sim": 0.9923640191555023, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007635983452200889, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.16924743652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9520233154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29850, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2005651980638504, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.1315156824886799, + "grad_norm": 0.0410173125565052, + "kept_token_count": 654.3625, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0068, + "mgd_cos_sim": 0.9932300269603729, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.006769977416843176, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.15723571777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7185791015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29860, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.20028364956378936, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.12324422895908356, + "grad_norm": 0.04741079732775688, + "kept_token_count": 654.2578125, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9926639080047608, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007336088921874762, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5345764160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2925628662109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29870, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19893933832645416, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.12518348321318626, + "grad_norm": 0.05245641618967056, + "kept_token_count": 655.3875, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0073, + "mgd_cos_sim": 0.9926532685756684, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007346729096025228, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7141387939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5899627685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29880, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.1991215854883194, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.12833005785942078, + "grad_norm": 0.04091213271021843, + "kept_token_count": 655.28125, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0071, + "mgd_cos_sim": 0.9928563475608826, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007143652532249689, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.18870849609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7564666748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29890, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20031485259532927, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.12602993622422218, + "grad_norm": 0.04878632351756096, + "kept_token_count": 654.5453125, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0076, + "mgd_cos_sim": 0.9923919022083283, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007608095183968544, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3067932128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.25958251953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29900, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.1998957946896553, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.12528123036026956, + "grad_norm": 0.039120785892009735, + "kept_token_count": 654.8734375, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0074, + "mgd_cos_sim": 0.9926191091537475, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007380891498178244, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.01915283203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.53313903808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29910, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.19949934631586075, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.12390049546957016, + "grad_norm": 0.03724857419729233, + "kept_token_count": 654.90625, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0074, + "mgd_cos_sim": 0.9925786077976226, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007421396858990193, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.73743591308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.9888031005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29920, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2003055766224861, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.12876341566443444, + "grad_norm": 0.061121538281440735, + "kept_token_count": 654.25625, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0082, + "mgd_cos_sim": 0.9918266355991363, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.008173358719795942, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8054931640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.11220397949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29930, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.1997283071279526, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.13302449733018876, + "grad_norm": 0.0602872259914875, + "kept_token_count": 654.8578125, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0075, + "mgd_cos_sim": 0.9925388038158417, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007461198326200247, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.83837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7103973388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29940, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20115787386894227, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.1283969447016716, + "grad_norm": 0.04773528501391411, + "kept_token_count": 653.678125, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0077, + "mgd_cos_sim": 0.9922875046730042, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007712492160499096, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2854431152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.63857116699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19927092790603637, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.13689747974276542, + "grad_norm": 0.045590322464704514, + "kept_token_count": 655.19375, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0076, + "mgd_cos_sim": 0.9924043118953705, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007595689035952091, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6256164550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5654724121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29960, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20054861158132553, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.13388489484786986, + "grad_norm": 0.05107124522328377, + "kept_token_count": 653.9953125, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0072, + "mgd_cos_sim": 0.9927838563919067, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007216154877096414, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.04681091308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6568634033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29970, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.19972842633724214, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.1300174981355667, + "grad_norm": 0.04976195469498634, + "kept_token_count": 654.6171875, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0074, + "mgd_cos_sim": 0.9925871729850769, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007412820681929588, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0470672607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4075073242187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29980, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.20019918382167817, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.12049296721816063, + "grad_norm": 0.04404812306165695, + "kept_token_count": 654.59375, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0071, + "mgd_cos_sim": 0.9929125070571899, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007087483908981085, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6448608398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.13731384277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29990, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.2011650964617729, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.12422739341855049, + "grad_norm": 0.042815446853637695, + "kept_token_count": 653.89375, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0073, + "mgd_cos_sim": 0.9927427530288696, + "mgd_fm_loss_weight": 0.0, + "mgd_loss": 0.007257253583520651, + "mgd_loss_weight": 1.0, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7632659912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.78380126953124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 30000, + "valid_token_count": 818.5625 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.020082239318142334, + "train_runtime": 49897.7616, + "train_samples_per_second": 76.957, + "train_steps_per_second": 0.601 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b4007459e3fe5714b85aae807cd7620bee3a8d36 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase2/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c71cb09268e8e60d2ffe83726e03428cb07ed1757530c8243d93fcd0b079f5b2 +size 5969 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/config.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..40634fb646227d5419d3f5fbde7d050d43eb48d1 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": true, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.5, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": null, + "mgd_loss_weight_start": 0.05, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.2, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00001-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..53daa36095f758349e079472f83566a64083062f --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52cec7dc7d09460862211807a09deaa1b1004e9c594a0fee41f23b1779a0db23 +size 4999367032 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00002-of-00002.safetensors b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cd3210ab64cdc751111bd44e08e99171103d41c4 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:554bba1a1e55dfc1eac9c7627c16ab3f4af1e402711a854dcd84b028ea3a4de5 +size 2643339748 diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model.safetensors.index.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/trainer_state.json b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..af565b613b1bd054bf920d32aa1a78ca4284a55f --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.19896580427885055, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.14447055608034134, + "grad_norm": 0.8678640723228455, + "kept_token_count": 655.6640625, + "learning_rate": 1.8e-07, + "loss": 0.1486, + "mgd_cos_sim": 0.9918397724628448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00816024299710989, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.06885986328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.85560607910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.2001817539334297, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.13655784949660302, + "grad_norm": 0.9702594876289368, + "kept_token_count": 654.5171875, + "learning_rate": 3.8e-07, + "loss": 0.1402, + "mgd_cos_sim": 0.9926866352558136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007313355337828398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.31101989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1316833496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.1994730532169342, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.12430142387747764, + "grad_norm": 0.8428348302841187, + "kept_token_count": 655.1265625, + "learning_rate": 5.800000000000001e-07, + "loss": 0.128, + "mgd_cos_sim": 0.9926602482795716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007339748553931713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.22776184082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6907012939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 30, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19924101382493972, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.10124831050634384, + "grad_norm": 0.6334657669067383, + "kept_token_count": 655.4390625, + "learning_rate": 7.799999999999999e-07, + "loss": 0.1049, + "mgd_cos_sim": 0.9926814794540405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007318521197885275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.31886901855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9224090576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 40, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.19993981570005417, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.08988177478313446, + "grad_norm": 0.5420099496841431, + "kept_token_count": 654.6515625, + "learning_rate": 9.8e-07, + "loss": 0.0935, + "mgd_cos_sim": 0.9928047358989716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00719525022432208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.49432067871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2146331787109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 50, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20070015341043473, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.0736708927899599, + "grad_norm": 0.49591147899627686, + "kept_token_count": 653.9328125, + "learning_rate": 1.18e-06, + "loss": 0.0777, + "mgd_cos_sim": 0.9920114636421203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007988538965582848, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.57334594726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.65673828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 60, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20017197728157043, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.06346535235643387, + "grad_norm": 0.585085391998291, + "kept_token_count": 654.5875, + "learning_rate": 1.38e-06, + "loss": 0.0674, + "mgd_cos_sim": 0.992121285200119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007878698874264956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.74169006347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2034057617187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 70, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.19966575056314467, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.04687256775796413, + "grad_norm": 0.3265725076198578, + "kept_token_count": 654.909375, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0506, + "mgd_cos_sim": 0.9926087856292725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007391209248453379, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.23598327636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.9772888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 80, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19937237352132797, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.04559928886592388, + "grad_norm": 0.35714876651763916, + "kept_token_count": 655.003125, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0499, + "mgd_cos_sim": 0.9913111805915833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008688811585307122, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.5639404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.48651733398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 90, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20029806792736055, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.038097736239433286, + "grad_norm": 0.1931251734495163, + "kept_token_count": 654.4546875, + "learning_rate": 1.98e-06, + "loss": 0.0422, + "mgd_cos_sim": 0.9917385995388031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261400740593672, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.89795837402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.90501403808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 100, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.2003684461116791, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.03967490177601576, + "grad_norm": 0.19658222794532776, + "kept_token_count": 654.16875, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0439, + "mgd_cos_sim": 0.9915131866931916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00848680892959237, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.7414581298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5901702880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 110, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2004957988858223, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.033236847072839734, + "grad_norm": 0.18801656365394592, + "kept_token_count": 654.13125, + "learning_rate": 2.38e-06, + "loss": 0.0376, + "mgd_cos_sim": 0.9913115680217743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008688432816416024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.70659790039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.72874145507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 120, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19953783601522446, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.030173953250050546, + "grad_norm": 0.2010108381509781, + "kept_token_count": 654.6953125, + "learning_rate": 2.58e-06, + "loss": 0.0344, + "mgd_cos_sim": 0.9914868474006653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008513152226805687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.54625854492187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.78692626953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 130, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.1992308035492897, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.031097463518381118, + "grad_norm": 0.22844745218753815, + "kept_token_count": 655.3046875, + "learning_rate": 2.78e-06, + "loss": 0.0354, + "mgd_cos_sim": 0.9913648843765259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008635111805051564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.3307098388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8898223876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2004428654909134, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.031114835292100906, + "grad_norm": 0.21278877556324005, + "kept_token_count": 654.1921875, + "learning_rate": 2.98e-06, + "loss": 0.0354, + "mgd_cos_sim": 0.9914456486701966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008554366417229176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.15827178955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2163330078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 150, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20019686073064805, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.02739346120506525, + "grad_norm": 0.2656020224094391, + "kept_token_count": 654.5125, + "learning_rate": 3.18e-06, + "loss": 0.0321, + "mgd_cos_sim": 0.9905736565589904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009426351916044951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.62579650878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.35504455566405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 160, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2001825526356697, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.02374684512615204, + "grad_norm": 0.15878033638000488, + "kept_token_count": 654.234375, + "learning_rate": 3.38e-06, + "loss": 0.0285, + "mgd_cos_sim": 0.9904101610183715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00958984699100256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.33515319824218, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0338531494141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.2010756954550743, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.021655075252056122, + "grad_norm": 0.1995282918214798, + "kept_token_count": 653.8625, + "learning_rate": 3.58e-06, + "loss": 0.0262, + "mgd_cos_sim": 0.9909537017345429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009046295844018459, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.2915481567383, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0220886230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 180, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2004741206765175, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.021900398284196855, + "grad_norm": 0.1876661479473114, + "kept_token_count": 654.3859375, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0266, + "mgd_cos_sim": 0.9905412018299102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009458801336586475, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.17205505371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2038513183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 190, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.1996339201927185, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.022286617942154408, + "grad_norm": 0.16695930063724518, + "kept_token_count": 654.8609375, + "learning_rate": 3.98e-06, + "loss": 0.0272, + "mgd_cos_sim": 0.9901743650436401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009825639333575964, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.9789596557617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.30584106445315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 200, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19937500804662706, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.022061285469681023, + "grad_norm": 0.14283892512321472, + "kept_token_count": 655.396875, + "learning_rate": 4.18e-06, + "loss": 0.0274, + "mgd_cos_sim": 0.98933624625206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010663748253136873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 164.9552474975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.26927185058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 210, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.20005813091993332, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.020730129815638064, + "grad_norm": 0.1801074594259262, + "kept_token_count": 654.2265625, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0258, + "mgd_cos_sim": 0.9899081468582154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010091866739094258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.43832244873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1520202636719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 220, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.19934505969285965, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.023311034962534904, + "grad_norm": 0.14708219468593597, + "kept_token_count": 654.86875, + "learning_rate": 4.58e-06, + "loss": 0.0285, + "mgd_cos_sim": 0.9895709156990051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010429079737514257, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.49846496582032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6298126220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 230, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.19903492629528047, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.02491577174514532, + "grad_norm": 0.14959383010864258, + "kept_token_count": 655.1625, + "learning_rate": 4.78e-06, + "loss": 0.03, + "mgd_cos_sim": 0.989928525686264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010071467608213425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.098828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8258361816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 240, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.19959159642457963, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.02069023884832859, + "grad_norm": 0.13413359224796295, + "kept_token_count": 655.18125, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0258, + "mgd_cos_sim": 0.9898180544376374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010181946773082018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.08339080810546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7520355224609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 250, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.20096953064203263, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.019264704920351504, + "grad_norm": 0.1419239491224289, + "kept_token_count": 653.8421875, + "learning_rate": 5.18e-06, + "loss": 0.0241, + "mgd_cos_sim": 0.9903787434101105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00962124215438962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 156.7454864501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.53697814941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 260, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19994893819093704, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.020439049508422614, + "grad_norm": 0.12378499656915665, + "kept_token_count": 654.5296875, + "learning_rate": 5.38e-06, + "loss": 0.0256, + "mgd_cos_sim": 0.989579153060913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010420838650316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 148.37352142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.66115112304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 270, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20005052536725998, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.01838732771575451, + "grad_norm": 0.13127462565898895, + "kept_token_count": 654.4453125, + "learning_rate": 5.58e-06, + "loss": 0.0232, + "mgd_cos_sim": 0.990317577123642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009682435356080533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.45783081054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67267150878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 280, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20024654567241668, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.017940921429544687, + "grad_norm": 0.13233016431331635, + "kept_token_count": 654.296875, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0227, + "mgd_cos_sim": 0.9904550611972809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009544937591999769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.8318649291992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.48316650390626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 290, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2005375489592552, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.021161306835711, + "grad_norm": 0.12514306604862213, + "kept_token_count": 653.9484375, + "learning_rate": 5.98e-06, + "loss": 0.0263, + "mgd_cos_sim": 0.9896327435970307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010367259196937085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 147.04932861328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2814239501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 300, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.1995272159576416, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.02163562849164009, + "grad_norm": 0.10456936806440353, + "kept_token_count": 655.0578125, + "learning_rate": 6.18e-06, + "loss": 0.0265, + "mgd_cos_sim": 0.9902462184429168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009753784723579883, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.84363555908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6503143310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19961801618337632, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.01599944168701768, + "grad_norm": 0.1305389404296875, + "kept_token_count": 654.8859375, + "learning_rate": 6.38e-06, + "loss": 0.0206, + "mgd_cos_sim": 0.9907208859920502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009279116056859493, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.04188537597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 320, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20063386708498002, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.0204979763366282, + "grad_norm": 0.13593745231628418, + "kept_token_count": 653.9796875, + "learning_rate": 6.58e-06, + "loss": 0.0257, + "mgd_cos_sim": 0.9896627604961395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01033723894506693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.22388916015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.40582885742185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 330, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19990057945251466, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.023326240479946136, + "grad_norm": 0.14328983426094055, + "kept_token_count": 654.7328125, + "learning_rate": 6.78e-06, + "loss": 0.0284, + "mgd_cos_sim": 0.9897957861423492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01020421925932169, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 143.1925491333008, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.4170257568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 340, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.19963017255067825, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.018937612138688564, + "grad_norm": 0.13647592067718506, + "kept_token_count": 654.6859375, + "learning_rate": 6.98e-06, + "loss": 0.0237, + "mgd_cos_sim": 0.99055016040802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009449833538383245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.30272369384767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.92552795410154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 350, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19952460676431655, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.021477737557142974, + "grad_norm": 0.11813528090715408, + "kept_token_count": 655.2125, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0263, + "mgd_cos_sim": 0.9903818070888519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00961820175871253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.69305572509765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.69544982910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 360, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.20073167830705643, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.018356536235660314, + "grad_norm": 0.10218677669763565, + "kept_token_count": 654.1515625, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0234, + "mgd_cos_sim": 0.9898545682430268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010145439207553864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 144.48135681152343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8658721923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 370, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20114568322896959, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.017964962124824523, + "grad_norm": 0.10983534902334213, + "kept_token_count": 653.625, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0233, + "mgd_cos_sim": 0.9892415285110474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010758465621620417, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.72484436035157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.67969665527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 380, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19991480857133864, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.017162487842142582, + "grad_norm": 0.09862276911735535, + "kept_token_count": 654.5453125, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0222, + "mgd_cos_sim": 0.9899012684822083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010098745673894882, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 150.76910400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.62395935058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 390, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2005770057439804, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.018073005601763725, + "grad_norm": 0.11489212512969971, + "kept_token_count": 654.0890625, + "learning_rate": 7.98e-06, + "loss": 0.0228, + "mgd_cos_sim": 0.9905117332935334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009488263633102179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.58226623535157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.980322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 400, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20035794079303743, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.019491209648549557, + "grad_norm": 0.09637714177370071, + "kept_token_count": 654.1359375, + "learning_rate": 8.18e-06, + "loss": 0.0242, + "mgd_cos_sim": 0.9906562685966491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009343726187944412, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.95427398681642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2159393310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 410, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.19934803247451782, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.017838383745402097, + "grad_norm": 0.0924602746963501, + "kept_token_count": 655.1625, + "learning_rate": 8.38e-06, + "loss": 0.0226, + "mgd_cos_sim": 0.9903969764709473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0096030343323946, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.06877593994142, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.71880798339845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 420, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19985261112451552, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.0210218645632267, + "grad_norm": 0.14675858616828918, + "kept_token_count": 654.6515625, + "learning_rate": 8.58e-06, + "loss": 0.0264, + "mgd_cos_sim": 0.9892598092556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01074016997590661, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.42462005615235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.22727966308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20123629420995712, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.01806046925485134, + "grad_norm": 0.11686249077320099, + "kept_token_count": 653.840625, + "learning_rate": 8.78e-06, + "loss": 0.0229, + "mgd_cos_sim": 0.9903208374977112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00967916939407587, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.3919692993164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.25893249511716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 440, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.1992733284831047, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.016930181346833707, + "grad_norm": 0.11453676968812943, + "kept_token_count": 655.11875, + "learning_rate": 8.98e-06, + "loss": 0.0215, + "mgd_cos_sim": 0.9909401595592499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009059834107756615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 149.71383819580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4117126464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20101766884326935, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.016629189997911454, + "grad_norm": 0.11722355335950851, + "kept_token_count": 653.771875, + "learning_rate": 9.18e-06, + "loss": 0.0213, + "mgd_cos_sim": 0.9905965685844421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009403428807854653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 152.67705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.85435485839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 460, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19942198544740677, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.016589907836169003, + "grad_norm": 0.10838896781206131, + "kept_token_count": 655.1453125, + "learning_rate": 9.38e-06, + "loss": 0.0215, + "mgd_cos_sim": 0.990210485458374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00978951882570982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 151.49664764404298, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8963348388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 470, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19899210780858995, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.01807855237275362, + "grad_norm": 0.14087142050266266, + "kept_token_count": 655.56875, + "learning_rate": 9.580000000000001e-06, + "loss": 0.023, + "mgd_cos_sim": 0.9900933980941773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009906608797609805, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.75900421142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.7223815917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 480, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.19999441653490066, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.017383658699691294, + "grad_norm": 0.129452183842659, + "kept_token_count": 654.721875, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0223, + "mgd_cos_sim": 0.9902287483215332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009771246556192636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.5616455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.56787109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 490, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19975856095552444, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.016913417167961596, + "grad_norm": 0.10343674570322037, + "kept_token_count": 654.575, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0215, + "mgd_cos_sim": 0.9908206403255463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009179354924708606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 155.42460174560546, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.8479034423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 500, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20165106952190398, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.01489348877221346, + "grad_norm": 0.12274976819753647, + "kept_token_count": 653.2109375, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9907638311386109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009236192610114814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 162.01845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.8876708984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 510, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.19997855871915818, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.01821936424821615, + "grad_norm": 0.13260817527770996, + "kept_token_count": 654.71875, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9905609607696533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009439037833362817, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.8237060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.10169372558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 520, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20015983879566193, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.017842315882444382, + "grad_norm": 0.14818046987056732, + "kept_token_count": 654.43125, + "learning_rate": 1.058e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9904153406620025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009584664646536113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 154.60875091552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.863037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 530, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19958849102258683, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.01947212256491184, + "grad_norm": 0.12909947335720062, + "kept_token_count": 654.8796875, + "learning_rate": 1.078e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9902725994586945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00972739877179265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.79579010009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7351867675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 540, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19952405542135238, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.01545304274186492, + "grad_norm": 0.11819788068532944, + "kept_token_count": 655.221875, + "learning_rate": 1.098e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9896706938743591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010329298675060272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.72737884521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.2541809082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 550, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.19956788569688796, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.019252650905400515, + "grad_norm": 0.13040754199028015, + "kept_token_count": 654.809375, + "learning_rate": 1.118e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9905770778656006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009422925673425198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.31761932373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.6694702148437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 560, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.1994061663746834, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.01844470826908946, + "grad_norm": 0.12307389825582504, + "kept_token_count": 655.05, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9914593756198883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008540613856166602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.70960693359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6387939453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 570, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19936722218990327, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.017762023862451314, + "grad_norm": 0.1389349400997162, + "kept_token_count": 654.9453125, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9910053253173828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008994680922478437, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 153.52964935302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2574798583984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 580, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.19999982714653014, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.01776466267183423, + "grad_norm": 0.1061241626739502, + "kept_token_count": 654.6046875, + "learning_rate": 1.178e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9903080284595489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009691974520683289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.3802505493164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5774169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 590, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19993949234485625, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.017943057511001826, + "grad_norm": 0.10731880366802216, + "kept_token_count": 654.7328125, + "learning_rate": 1.198e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9911051452159881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008894868567585945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.25506439208985, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.94244995117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 600, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19969477653503417, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.0165685560554266, + "grad_norm": 0.1265622079372406, + "kept_token_count": 654.753125, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9906563401222229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009343668073415756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 157.55730285644532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 355.7610198974609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 610, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20051750540733337, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.016886445600539444, + "grad_norm": 0.10254649072885513, + "kept_token_count": 654.0125, + "learning_rate": 1.238e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9906027972698211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009397203382104636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 159.93706512451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.50693969726564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 620, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.2009072259068489, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.018888500146567823, + "grad_norm": 0.08739899843931198, + "kept_token_count": 653.9609375, + "learning_rate": 1.258e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9902066290378571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009793368354439735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 160.26815795898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.2736480712891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 630, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2000364914536476, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.018535403348505498, + "grad_norm": 0.10777078568935394, + "kept_token_count": 654.5125, + "learning_rate": 1.278e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912085592746734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008791432064026595, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.7313262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3691345214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 640, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.20039894580841064, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.01661941222846508, + "grad_norm": 0.11734291911125183, + "kept_token_count": 654.16875, + "learning_rate": 1.298e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9914046704769135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008595329150557518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.1794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.115478515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 650, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20048571974039078, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.018214789032936097, + "grad_norm": 0.11551355570554733, + "kept_token_count": 654.1640625, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.991145807504654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008854183461517095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 163.7718933105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3473724365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 660, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19962664395570756, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.01690060831606388, + "grad_norm": 0.11907830089330673, + "kept_token_count": 655.0203125, + "learning_rate": 1.338e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9911172449588775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00888274684548378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.61397705078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6469329833984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 670, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1995181381702423, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.01626321831718087, + "grad_norm": 0.13202816247940063, + "kept_token_count": 654.9578125, + "learning_rate": 1.358e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9910962522029877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008903750590980053, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 169.74379119873046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.1880218505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 680, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19988448321819305, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.014584302343428135, + "grad_norm": 0.09705588221549988, + "kept_token_count": 654.7828125, + "learning_rate": 1.378e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9913655519485474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008634443953633308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 170.97698974609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3138397216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 690, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20040232837200164, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.01681305021047592, + "grad_norm": 0.1417911946773529, + "kept_token_count": 654.484375, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9910525560379029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008947428781539202, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 165.02399597167968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.981396484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 700, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19917096346616744, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.017071035876870155, + "grad_norm": 0.1131586804986, + "kept_token_count": 655.5625, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9907368361949921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00926316436380148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 167.2938018798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.4125701904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 710, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.20065061002969742, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.0162603285163641, + "grad_norm": 0.0926605612039566, + "kept_token_count": 653.8828125, + "learning_rate": 1.438e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9905286312103272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009471358079463244, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.3166488647461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.57826232910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 720, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.19956629276275634, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.016331125982105732, + "grad_norm": 0.09670574218034744, + "kept_token_count": 655.034375, + "learning_rate": 1.458e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9910029947757721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008996997401118279, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.77726287841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.6235595703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 730, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2005702883005142, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.018585965875536202, + "grad_norm": 0.1133778989315033, + "kept_token_count": 654.1015625, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9908369958400727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009162992984056473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.15591278076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.90581970214845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 740, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19925353825092315, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.018509041611105205, + "grad_norm": 0.14687217772006989, + "kept_token_count": 655.3265625, + "learning_rate": 1.498e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.990748155117035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00925183417275548, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.01067962646485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.5264190673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 750, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.19966906309127808, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.017731878999620677, + "grad_norm": 0.15428155660629272, + "kept_token_count": 654.7734375, + "learning_rate": 1.518e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9914759159088135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008524088002741338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.65125122070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0429748535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 760, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.19965673089027405, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.020124489348381756, + "grad_norm": 0.15275904536247253, + "kept_token_count": 654.7203125, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.99065962433815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009340375196188688, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 171.59982604980468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9485229492187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 770, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.2000300720334053, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.01719945864751935, + "grad_norm": 0.11163019388914108, + "kept_token_count": 654.4390625, + "learning_rate": 1.558e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9912121593952179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008787844143807889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.66925506591798, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6775726318359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 780, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.199848672747612, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.015323039889335633, + "grad_norm": 0.1257646530866623, + "kept_token_count": 654.3703125, + "learning_rate": 1.578e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9915585219860077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008441490679979324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.95841369628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4126403808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 790, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.19984670728445053, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.017693113163113595, + "grad_norm": 0.1258036494255066, + "kept_token_count": 654.4890625, + "learning_rate": 1.598e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9909364938735962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009063487127423286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.3444046020508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0159912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 800, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.19936944544315338, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.014035457652062178, + "grad_norm": 0.10165678709745407, + "kept_token_count": 655.053125, + "learning_rate": 1.618e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9909823179244995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009017689898610116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.94364013671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.56313171386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 810, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.20056470781564711, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.01633323561400175, + "grad_norm": 0.11945604532957077, + "kept_token_count": 653.9953125, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9911266982555389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008873309381306172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 173.79255065917968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2916198730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 820, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20042721629142762, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.01606285395100713, + "grad_norm": 0.11102742701768875, + "kept_token_count": 654.0125, + "learning_rate": 1.658e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.991191053390503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008808954246342182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.53568115234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7954071044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 830, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.20022043138742446, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.01823140885680914, + "grad_norm": 0.11991652101278305, + "kept_token_count": 654.4640625, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9910285294055938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008971471339464188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.94774322509767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7950469970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 840, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.19946259409189224, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.018631458934396505, + "grad_norm": 0.11873369663953781, + "kept_token_count": 654.99375, + "learning_rate": 1.698e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903041779994964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009695825167000294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 176.06033782958986, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0024780273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 850, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19913458675146103, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.017703148163855077, + "grad_norm": 0.12582874298095703, + "kept_token_count": 655.5578125, + "learning_rate": 1.718e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9909742057323456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009025780390948058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.81346588134767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.37470703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 860, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.2006307139992714, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.017714133858680724, + "grad_norm": 0.12636113166809082, + "kept_token_count": 653.821875, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9899966716766357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010003331117331982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.08179168701173, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2841796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 870, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.20049131959676741, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.018450725823640823, + "grad_norm": 0.1082620918750763, + "kept_token_count": 654.434375, + "learning_rate": 1.758e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9904475152492523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009552469570189714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.75672454833983, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2209808349609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 880, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.1998995140194893, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.016490449011325837, + "grad_norm": 0.12235110998153687, + "kept_token_count": 654.5, + "learning_rate": 1.778e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9911931276321411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008806864637881518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.21283569335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7047882080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 890, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2003964051604271, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.018521070200949907, + "grad_norm": 0.09188554435968399, + "kept_token_count": 654.2109375, + "learning_rate": 1.798e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9909106492996216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009089353028684855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 172.41013946533204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.3441497802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 900, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.20054490268230438, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.017450337205082177, + "grad_norm": 0.10819627344608307, + "kept_token_count": 654.0171875, + "learning_rate": 1.818e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9904263138771057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009573682863265275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 177.32910766601563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.35524291992186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 910, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.1993338331580162, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.015747595205903055, + "grad_norm": 0.1255778670310974, + "kept_token_count": 655.2546875, + "learning_rate": 1.838e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9909935772418976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900642741471529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.87675933837892, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.14251708984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 920, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.20000912845134736, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.018745314329862595, + "grad_norm": 0.10835038125514984, + "kept_token_count": 654.7296875, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9906004965305328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009399514645338058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.47678985595704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1649505615234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 930, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20033377707004546, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.017463009525090455, + "grad_norm": 0.10556039214134216, + "kept_token_count": 654.3453125, + "learning_rate": 1.878e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9899864196777344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010013577062636614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.5413833618164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7145477294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 940, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1991698145866394, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.016357979737222195, + "grad_norm": 0.1285199224948883, + "kept_token_count": 655.61875, + "learning_rate": 1.898e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9913159012794495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008684078045189381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.05094451904296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.3283935546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 950, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.2001700446009636, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.01745589738711715, + "grad_norm": 0.13919128477573395, + "kept_token_count": 654.709375, + "learning_rate": 1.918e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.991135448217392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008864532876759767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 175.0017074584961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.57837829589846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 960, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.20102721452713013, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.017623960226774215, + "grad_norm": 0.12866871058940887, + "kept_token_count": 653.778125, + "learning_rate": 1.938e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9906578958034515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009342113882303238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.015771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8447631835937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20072255581617354, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.014515822380781173, + "grad_norm": 0.10161425918340683, + "kept_token_count": 653.871875, + "learning_rate": 1.958e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9917542338371277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00824576448649168, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.47572174072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7182342529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 980, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.1997675985097885, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.02132464023306966, + "grad_norm": 0.1299259066581726, + "kept_token_count": 655.0875, + "learning_rate": 1.978e-05, + "loss": 0.026, + "mgd_cos_sim": 0.990692549943924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009307452570647001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.58450622558593, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.34877319335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 990, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.19930484294891357, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.016337068285793066, + "grad_norm": 0.14119897782802582, + "kept_token_count": 655.2453125, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9908827781677246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009117225650697947, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.43663177490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7796997070312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1000, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19965189695358276, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.01709466828033328, + "grad_norm": 0.13431403040885925, + "kept_token_count": 654.9046875, + "learning_rate": 2.018e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9905260503292084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00947394985705614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.91541290283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.11370239257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1010, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.19981635361909866, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.021076018922030926, + "grad_norm": 0.16058778762817383, + "kept_token_count": 654.715625, + "learning_rate": 2.038e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9904288828372956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009571107663214206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 190.29697875976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.79526672363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.20054191201925278, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.016575275268405677, + "grad_norm": 0.1472502201795578, + "kept_token_count": 654.3125, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9909491121768952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009050896298140288, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.26629180908202, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 354.8921356201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1030, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.19935457706451415, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.015302198566496372, + "grad_norm": 0.11910255253314972, + "kept_token_count": 655.253125, + "learning_rate": 2.078e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9902119696140289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009788030479103326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 185.1519989013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.0312805175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1040, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.2005959153175354, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.016507707070559263, + "grad_norm": 0.13078522682189941, + "kept_token_count": 654.4578125, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.99064199924469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009358008857816458, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 191.36114807128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7743377685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1050, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.20052432715892793, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.014466156158596278, + "grad_norm": 0.12107322365045547, + "kept_token_count": 654.075, + "learning_rate": 2.118e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9907048761844635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009295127168297768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 189.26429595947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.0950531005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1060, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20081171989440919, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.017220241855829953, + "grad_norm": 0.1501845121383667, + "kept_token_count": 654.0953125, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9903954029083252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00960458256304264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.74794311523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.44182739257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1070, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.20032891184091567, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.01863296953961253, + "grad_norm": 0.15000097453594208, + "kept_token_count": 654.40625, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9906122922897339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009387707337737083, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.27681579589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0253509521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1080, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.1992271661758423, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.015878020599484444, + "grad_norm": 0.13890162110328674, + "kept_token_count": 655.2234375, + "learning_rate": 2.178e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9913037240505218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008696278743445873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.96995849609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5136474609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1090, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20019873827695847, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.016126709431409834, + "grad_norm": 0.10474548488855362, + "kept_token_count": 654.403125, + "learning_rate": 2.198e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9909402251243591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009059769473969937, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.88306274414063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0945098876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1100, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.19982388615608215, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.015252935234457254, + "grad_norm": 0.13644076883792877, + "kept_token_count": 654.8046875, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9904405415058136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009559458121657372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.92591552734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.59510192871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1110, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.19920990169048308, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.015412667114287616, + "grad_norm": 0.08684926480054855, + "kept_token_count": 655.3796875, + "learning_rate": 2.238e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9909601211547852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009039879590272904, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.9782928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7907958984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1120, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19914452880620956, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.017952363565564157, + "grad_norm": 0.1374688595533371, + "kept_token_count": 655.409375, + "learning_rate": 2.258e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9910044729709625, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008995533362030983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 210.6736312866211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0243774414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1130, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.19945166409015655, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.01818876750767231, + "grad_norm": 0.13581885397434235, + "kept_token_count": 655.2671875, + "learning_rate": 2.278e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.990236473083496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009763513691723347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.9071060180664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3761749267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1140, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.20004442632198333, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.01461706990376115, + "grad_norm": 0.11055659502744675, + "kept_token_count": 654.4546875, + "learning_rate": 2.298e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9911613047122956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008838693052530289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.73024139404296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.50537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1150, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.2005326122045517, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.018006753455847502, + "grad_norm": 0.13190992176532745, + "kept_token_count": 654.209375, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9907960295677185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009203969966620207, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.0991989135742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.7565887451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1160, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2005370631814003, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.01937542874366045, + "grad_norm": 0.13322731852531433, + "kept_token_count": 653.9140625, + "learning_rate": 2.338e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9904799163341522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009520077146589756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 198.86636810302736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9096313476563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1170, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.19990890771150588, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.017958430852741004, + "grad_norm": 0.13456568121910095, + "kept_token_count": 654.8234375, + "learning_rate": 2.358e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9900281012058259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00997188724577427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.0243392944336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.14691467285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1180, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.20033340752124787, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.015563501976430416, + "grad_norm": 0.15841618180274963, + "kept_token_count": 654.4125, + "learning_rate": 2.378e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9906444728374482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00935553116723895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.26116333007812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8070556640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1190, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1988750621676445, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.018799261283129455, + "grad_norm": 0.16547921299934387, + "kept_token_count": 655.7125, + "learning_rate": 2.398e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9901657581329346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009834239073097707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 212.68112182617188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5520294189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1200, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19949506223201752, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.01580273825675249, + "grad_norm": 0.1056087464094162, + "kept_token_count": 654.74375, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9909103095531464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009089685697108506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.20691528320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.62550964355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1210, + "valid_token_count": 817.9078125 + }, + { + "actual_token_mask_ratio": 0.2003350406885147, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.01590552618727088, + "grad_norm": 0.20009687542915344, + "kept_token_count": 654.3265625, + "learning_rate": 2.438e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9907137751579285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009286222606897354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 212.64198150634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.13935546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1220, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19986653327941895, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.017706283181905747, + "grad_norm": 0.17096945643424988, + "kept_token_count": 654.3109375, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9904017508029938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009598251432180405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 201.3079071044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.87811279296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1230, + "valid_token_count": 817.75625 + }, + { + "actual_token_mask_ratio": 0.20086619704961778, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.016582028195261955, + "grad_norm": 0.12958276271820068, + "kept_token_count": 653.9, + "learning_rate": 2.478e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9907331943511963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009266810491681099, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.20206756591796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.99314880371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1240, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2003417432308197, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.016213405039161445, + "grad_norm": 0.251670241355896, + "kept_token_count": 654.2625, + "learning_rate": 2.498e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9905068933963775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009493117034435273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.20613555908204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7585388183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1250, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19863575249910354, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.016595491394400596, + "grad_norm": 0.11439899355173111, + "kept_token_count": 655.8453125, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.021, + "mgd_cos_sim": 0.991253787279129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008746208995580674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.54488525390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3395568847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1260, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.19982792437076569, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.015793257858604193, + "grad_norm": 0.11532047390937805, + "kept_token_count": 654.721875, + "learning_rate": 2.538e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9910733342170716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00892667742446065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.32331085205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.19585571289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1270, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20051511526107788, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.017646574601531028, + "grad_norm": 0.12088270485401154, + "kept_token_count": 654.209375, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9910467147827149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008953270502388477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.39821166992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.82810974121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1280, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19984420835971833, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.01724124690517783, + "grad_norm": 0.11104655265808105, + "kept_token_count": 654.45625, + "learning_rate": 2.578e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9908671855926514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009132813941687346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.17977447509764, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7290771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1290, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.2009379044175148, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.01654849210754037, + "grad_norm": 0.13560928404331207, + "kept_token_count": 653.7125, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9906300485134125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009369968064129352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.43277282714843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.00152893066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1300, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.20060714930295945, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.015198613982647657, + "grad_norm": 0.10164564102888107, + "kept_token_count": 654.175, + "learning_rate": 2.618e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9907573461532593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00924266055226326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.7959991455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2743286132812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1310, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.20001700669527053, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.01463364502415061, + "grad_norm": 0.1391582489013672, + "kept_token_count": 654.584375, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9906121909618377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009387800935655832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.8934753417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.49461059570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1320, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.199871788918972, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.019880064483731984, + "grad_norm": 0.12986783683300018, + "kept_token_count": 654.8609375, + "learning_rate": 2.658e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9898496150970459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010150372702628373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 204.61115417480468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.601904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1330, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2009448900818825, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.01841356148943305, + "grad_norm": 0.16256831586360931, + "kept_token_count": 654.1015625, + "learning_rate": 2.678e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.989620816707611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010379191022366285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.7258743286133, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 356.5174835205078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1340, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.2002723768353462, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.01679121991619468, + "grad_norm": 0.12076329439878464, + "kept_token_count": 654.4046875, + "learning_rate": 2.698e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9906561315059662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009343860018998385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 223.01512756347657, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4817352294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1350, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.1999894842505455, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.017257497552782296, + "grad_norm": 0.11575345695018768, + "kept_token_count": 654.5890625, + "learning_rate": 2.718e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9905914664268494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009408527798950673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.78343811035157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.93431091308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1360, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.1998303323984146, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.018287033960223197, + "grad_norm": 0.13990820944309235, + "kept_token_count": 654.8625, + "learning_rate": 2.738e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9902740120887756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009725980088114738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.3589630126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9500030517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1370, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.200156506896019, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.015474718064069748, + "grad_norm": 0.10033982247114182, + "kept_token_count": 654.6734375, + "learning_rate": 2.758e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9913089394569397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008691055234521627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 216.04017333984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.81833801269534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1380, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.19945200830698012, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.016932116448879243, + "grad_norm": 0.10691505670547485, + "kept_token_count": 654.959375, + "learning_rate": 2.778e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9904033005237579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009596697147935629, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.45682525634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.9456268310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1390, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19930353611707688, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.016337556671351193, + "grad_norm": 0.14814545214176178, + "kept_token_count": 655.3671875, + "learning_rate": 2.798e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.990801990032196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009198014624416828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.74014282226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.4509002685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1400, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.20003019720315934, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.016552389413118363, + "grad_norm": 0.22648240625858307, + "kept_token_count": 654.65, + "learning_rate": 2.818e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9894619464874268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010538059286773205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.16292572021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2063507080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1410, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.19951107501983642, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.017222218774259092, + "grad_norm": 0.10014734417200089, + "kept_token_count": 654.9375, + "learning_rate": 2.838e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9898306131362915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010169389005750418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.31914367675782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3553771972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1420, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19967983514070511, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.016086944192647935, + "grad_norm": 0.09499197453260422, + "kept_token_count": 654.7953125, + "learning_rate": 2.858e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9913212776184082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008678728248924017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.11069946289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5617919921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1430, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19977671802043914, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.01535549983382225, + "grad_norm": 0.11362940818071365, + "kept_token_count": 654.715625, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9905788004398346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009421194344758988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.61912994384767, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.404052734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1440, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2002604603767395, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.01585909379646182, + "grad_norm": 0.1208890974521637, + "kept_token_count": 654.45, + "learning_rate": 2.898e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9906916439533233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009308356884866953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.25709838867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.23069458007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1450, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20059887319803238, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.02118223924189806, + "grad_norm": 0.1404053419828415, + "kept_token_count": 653.9984375, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9898502588272095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010149731393903494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.27476959228517, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5456817626953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1460, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.20073195397853852, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.016721804905682803, + "grad_norm": 0.09345226734876633, + "kept_token_count": 654.059375, + "learning_rate": 2.938e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9904790937900543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00952089363709092, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 212.8689743041992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8491577148437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1470, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.1996222272515297, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.018719473201781512, + "grad_norm": 0.1145828440785408, + "kept_token_count": 655.0890625, + "learning_rate": 2.958e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903728485107421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009627163596451282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.14024505615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.81903381347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1480, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.1986422762274742, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.01775537319481373, + "grad_norm": 0.15370503067970276, + "kept_token_count": 655.6953125, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9904337644577026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009566233959048986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.0765869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7781158447266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1490, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20031036138534547, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.016943139769136906, + "grad_norm": 0.15431658923625946, + "kept_token_count": 654.5125, + "learning_rate": 2.998e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9901188611984253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0098811405710876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.39035186767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8163360595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1500, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.20052050054073334, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.017445919383317233, + "grad_norm": 0.11745694279670715, + "kept_token_count": 653.8234375, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9900704145431518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009929595328867436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.68865661621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.12880554199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1510, + "valid_token_count": 817.809375 + }, + { + "actual_token_mask_ratio": 0.19983989894390106, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.015306463092565536, + "grad_norm": 0.12959381937980652, + "kept_token_count": 654.65625, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9901950895786286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009804914239794016, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.5092788696289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.78758239746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1520, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20022741556167603, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.017654401250183582, + "grad_norm": 0.11287513375282288, + "kept_token_count": 654.621875, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908793568611145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009120640251785517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.59664154052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2799072265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1530, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.1995142802596092, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.018177666515111924, + "grad_norm": 0.1516772210597992, + "kept_token_count": 655.0640625, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906836092472077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009316406212747097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.16556243896486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1526153564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1540, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20082424134016036, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.018255690205842258, + "grad_norm": 0.12986372411251068, + "kept_token_count": 654.3109375, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9905939400196075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009406060818582774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.0355712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.06670532226565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1550, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.2000761643052101, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.01874489728361368, + "grad_norm": 0.10941766947507858, + "kept_token_count": 654.3875, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.990063464641571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009936530794948339, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.95906677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2309967041016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1560, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.20065777599811555, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.019456802029162647, + "grad_norm": 0.11662353575229645, + "kept_token_count": 654.1421875, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9901352643966674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009864730387926101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.54898376464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.41497802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1570, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19996454417705536, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.017382957879453896, + "grad_norm": 0.14303745329380035, + "kept_token_count": 654.4640625, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9898863911628724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01011360315605998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 232.88571166992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0251403808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1580, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.20030222982168197, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.01808843221515417, + "grad_norm": 0.11505771428346634, + "kept_token_count": 654.5453125, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9896777331829071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010322271380573511, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 228.20452575683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.64210205078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1590, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.1993904232978821, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.01883423337712884, + "grad_norm": 0.11252287030220032, + "kept_token_count": 655.1875, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9903862118721009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009613807685673237, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.27076873779296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9583251953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1600, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.1998190999031067, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.017938938830047844, + "grad_norm": 0.11181186884641647, + "kept_token_count": 654.7515625, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9896832346916199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010316753666847945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.6682327270508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84649047851565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1610, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19963840246200562, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.01616782434284687, + "grad_norm": 0.12734147906303406, + "kept_token_count": 654.625, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9901670813560486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009832912683486938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.4876937866211, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4431610107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1620, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.19985565841197966, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.020426703989505766, + "grad_norm": 0.1300031542778015, + "kept_token_count": 654.5625, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9898298919200897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010170100815594196, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.12533111572264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5653869628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.20086284279823302, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.01607084320858121, + "grad_norm": 0.1496485322713852, + "kept_token_count": 653.984375, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9900939404964447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009906070679426194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.86345825195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.49273376464845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20063425451517106, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.01737628160044551, + "grad_norm": 0.12352489680051804, + "kept_token_count": 653.9921875, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9894103527069091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010589650925248861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.5769775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3774658203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.20010327398777009, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.01580549944192171, + "grad_norm": 0.11282896995544434, + "kept_token_count": 654.5109375, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9903311371803284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009668860211968422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.91173248291017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2385498046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1660, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20074851512908937, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.018685822561383247, + "grad_norm": 0.1173023208975792, + "kept_token_count": 654.2859375, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9904531061649322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009546900540590287, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.0313690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5587982177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1670, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.19932288229465484, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.01871421029791236, + "grad_norm": 0.1623288244009018, + "kept_token_count": 655.2296875, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9902926504611969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009707354474812746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.62710876464843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.9792419433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1680, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20074892044067383, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.02023014975711703, + "grad_norm": 0.14183825254440308, + "kept_token_count": 653.8875, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9900783002376556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009921709075570107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.62647247314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.49431762695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19931895434856414, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.020671775750815868, + "grad_norm": 0.11347652226686478, + "kept_token_count": 655.2171875, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9900244355201722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009975561406463385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.66754455566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8263275146484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.19972368329763412, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.01763020558282733, + "grad_norm": 0.13743740320205688, + "kept_token_count": 654.8109375, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9906484365463257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009351549483835698, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.9001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5027313232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1710, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19972750395536423, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.018086136132478715, + "grad_norm": 0.13711445033550262, + "kept_token_count": 654.715625, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9903221070766449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009677894320338965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.93719940185548, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.34881896972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1720, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20059868693351746, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.01703840447589755, + "grad_norm": 0.13795718550682068, + "kept_token_count": 654.040625, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9901890873908996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009810920618474484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.0163146972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8837585449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1730, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19987414926290512, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.01715450091287494, + "grad_norm": 0.11222973465919495, + "kept_token_count": 654.73125, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9900071322917938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009992865566164254, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.46280670166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2759735107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1740, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19936608374118805, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.018842046707868577, + "grad_norm": 0.14293555915355682, + "kept_token_count": 655.14375, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9898734986782074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010126502625644207, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.3961944580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.22838439941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1750, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.200639571249485, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.016687577310949563, + "grad_norm": 0.1495954543352127, + "kept_token_count": 654.2359375, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9907767355442048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009223273023962975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.43282470703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.246533203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1760, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.19984454810619354, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.01962326532229781, + "grad_norm": 0.15750455856323242, + "kept_token_count": 654.490625, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9905764102935791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009423587936908007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.25527038574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6578857421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1770, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.1996798723936081, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.016825585346668958, + "grad_norm": 0.10410725325345993, + "kept_token_count": 654.915625, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9905319035053253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009468102362006903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.67357025146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8618225097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1780, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2001497983932495, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.01866651251912117, + "grad_norm": 0.16570289433002472, + "kept_token_count": 654.4109375, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9910654604434967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008934550639241933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.08871459960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8379211425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1790, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19993506520986556, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.018881043512374164, + "grad_norm": 0.15782028436660767, + "kept_token_count": 654.671875, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9895975351333618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010402473993599415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.49851989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4392974853516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1800, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20024954229593278, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.016399819683283568, + "grad_norm": 0.14660385251045227, + "kept_token_count": 654.3296875, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9901966392993927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009803359210491181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 233.8709487915039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.09166259765624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1810, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2002336248755455, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.017975838109850883, + "grad_norm": 0.12090631574392319, + "kept_token_count": 654.4421875, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9904879629611969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009512053336948156, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.28856811523437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0007568359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1820, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19960341453552247, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.016214730218052864, + "grad_norm": 0.12797749042510986, + "kept_token_count": 655.2328125, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9904941201210022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009505876060575246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.34808502197265, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6098297119141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1830, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20047151893377305, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.01998126143589616, + "grad_norm": 0.12010758370161057, + "kept_token_count": 654.30625, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9894955694675446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010504424944519997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.0407409667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.11268615722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1840, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2001402959227562, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.018041037395596503, + "grad_norm": 0.1406823843717575, + "kept_token_count": 654.3234375, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9903295814990998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00967042613774538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.23753967285157, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4823028564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1850, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20035559982061385, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.018821426946669818, + "grad_norm": 0.11596359312534332, + "kept_token_count": 654.1515625, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9898969054222106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010103089269250631, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 258.54402313232424, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.71881713867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1860, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20050826370716096, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.017297387309372424, + "grad_norm": 0.1013229638338089, + "kept_token_count": 654.24375, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9905360698699951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009463913086801768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.03623657226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.09959716796874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1870, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20101436227560043, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.017130862269550563, + "grad_norm": 0.1370367407798767, + "kept_token_count": 653.909375, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9905846297740937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00941537069156766, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.872412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.73939208984376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1880, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.199740406870842, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.019736220594495535, + "grad_norm": 0.14048774540424347, + "kept_token_count": 655.0078125, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.990576583147049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009423424769192935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 254.51967010498046, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6951416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1890, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.1995323807001114, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.017851654533296825, + "grad_norm": 0.1611417829990387, + "kept_token_count": 655.071875, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9900959610939026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009904021210968495, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.58014221191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7335571289062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1900, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19916568249464034, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.01792481793090701, + "grad_norm": 0.12385495007038116, + "kept_token_count": 655.3203125, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9896172225475312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010382784623652696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.9729751586914, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.2903289794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20034604370594025, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.01759271724149585, + "grad_norm": 0.12184280902147293, + "kept_token_count": 654.1921875, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9900316178798676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00996838603168726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.87916717529296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8190399169922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20060131698846817, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.017902069818228485, + "grad_norm": 0.11160140484571457, + "kept_token_count": 653.85, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9902299523353577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009770064521580935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.67189331054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.92694702148435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1930, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.20005988478660583, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.019172509387135505, + "grad_norm": 0.09470316767692566, + "kept_token_count": 654.578125, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9904711902141571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00952882245182991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.62747955322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.86615295410155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19913462400436402, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.017353366408497096, + "grad_norm": 0.1529383361339569, + "kept_token_count": 655.31875, + "learning_rate": 2.998163143144622e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9906618773937226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00933812502771616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.13869018554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4600067138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1950, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19947658479213715, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.016765029821544885, + "grad_norm": 0.12116070091724396, + "kept_token_count": 655.1046875, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9912027895450592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00879719341173768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.05877990722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.76727294921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1960, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.2004169449210167, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.01953022126108408, + "grad_norm": 0.16979657113552094, + "kept_token_count": 654.315625, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9900959074497223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009904089476913213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.9162124633789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6408752441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1970, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19978924244642257, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.0184085838496685, + "grad_norm": 0.1349037140607834, + "kept_token_count": 654.975, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9903179287910462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00968206999823451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.82415618896485, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6206298828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1980, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.19963604360818862, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.020064166374504568, + "grad_norm": 0.10435836017131805, + "kept_token_count": 654.9421875, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9895025551319122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010497442912310362, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.18187713623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6707305908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 1990, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.19931753128767013, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.016589562874287367, + "grad_norm": 0.12119099497795105, + "kept_token_count": 655.171875, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9902343213558197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009765668958425521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.04392700195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5992370605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2000, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19997284859418868, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.017644600104540588, + "grad_norm": 0.1335885226726532, + "kept_token_count": 654.4609375, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9906406283378602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009359370451420546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.9354019165039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2566345214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.20084254145622255, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.019313308876007795, + "grad_norm": 0.13260991871356964, + "kept_token_count": 653.859375, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9906121134757996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009387878980487586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.40710906982423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.81805725097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2020, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19920729845762253, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.018522934708744287, + "grad_norm": 0.16789944469928741, + "kept_token_count": 655.15, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9901347577571868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009865249879658223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 271.5802429199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4522644042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2030, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20011079013347627, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.017291317600756882, + "grad_norm": 0.12141552567481995, + "kept_token_count": 654.434375, + "learning_rate": 2.997353200422555e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9906101942062377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009389801230281592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.4524780273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7081268310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2040, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19916342943906784, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.019094123784452677, + "grad_norm": 0.13591112196445465, + "kept_token_count": 655.28125, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.990624338388443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009375671483576297, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.0853698730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0880096435547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20017271488904953, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.018305851705372335, + "grad_norm": 0.130568265914917, + "kept_token_count": 654.46875, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9904861807823181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00951381865888834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.6981170654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9587646484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2060, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20099021941423417, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.0204994460567832, + "grad_norm": 0.12226451933383942, + "kept_token_count": 653.68125, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.989713054895401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010286940541118384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.064306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.87669372558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2070, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20004552006721496, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.017146116960793732, + "grad_norm": 0.15821072459220886, + "kept_token_count": 654.63125, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.989569115638733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010430894792079926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.59612426757812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.61769409179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2080, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19939775317907332, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.019193488266319038, + "grad_norm": 0.13314497470855713, + "kept_token_count": 654.8875, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9906205475330353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009379460476338863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.3178771972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6167877197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2090, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.1988820031285286, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.0183435439132154, + "grad_norm": 0.13822682201862335, + "kept_token_count": 655.6859375, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9902844846248626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009715526551008224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.82359161376954, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5579071044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2100, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.2005731388926506, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.016342009231448174, + "grad_norm": 0.15400101244449615, + "kept_token_count": 654.209375, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9894127428531647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010587251279503107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.0576858520508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1812042236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2110, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.19985407292842866, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.01739087197929621, + "grad_norm": 0.15780693292617798, + "kept_token_count": 654.80625, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9906467854976654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009353217203170061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.9048675537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.41669921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2120, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20044710785150527, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.020902361162006854, + "grad_norm": 0.11985842883586884, + "kept_token_count": 654.3984375, + "learning_rate": 2.996395885442596e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9897867619991303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010213246382772923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.51267700195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.43898620605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2130, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.20062268823385238, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.016279454156756402, + "grad_norm": 0.14134198427200317, + "kept_token_count": 653.95625, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.990725040435791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009274945408105851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.3184341430664, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.72515869140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.19942691177129745, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.016887710336595772, + "grad_norm": 0.11840739101171494, + "kept_token_count": 655.225, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.990437364578247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009562618751078843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.217431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.43817138671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2150, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.2008420631289482, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.016501936223357917, + "grad_norm": 0.11660795658826828, + "kept_token_count": 654.0171875, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9903449475765228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009655049722641706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 250.65773162841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.0210479736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2160, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20008197724819182, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.015467970818281173, + "grad_norm": 0.16882339119911194, + "kept_token_count": 654.4203125, + "learning_rate": 2.995923131559128e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9909439504146575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009056052938103677, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.9562561035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4228973388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2170, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20008111894130706, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.017680493742227556, + "grad_norm": 0.13003113865852356, + "kept_token_count": 654.625, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9896503746509552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010349638666957617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.65155181884765, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.77594299316405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2180, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2008669912815094, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.016273416206240655, + "grad_norm": 0.12479858100414276, + "kept_token_count": 653.7421875, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9907134056091309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009286592435091733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.96197814941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.29888610839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20018212050199508, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.021470235008746386, + "grad_norm": 0.1735614836215973, + "kept_token_count": 654.678125, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.990817767381668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009182238765060902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.8983642578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.48094787597654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2200, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.1994966521859169, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.019309186656028033, + "grad_norm": 0.14025938510894775, + "kept_token_count": 655.284375, + "learning_rate": 2.995421294717254e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9907010018825531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00929900472983718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.0486801147461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8960815429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2210, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19979307949543, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.01920066848397255, + "grad_norm": 0.13052146136760712, + "kept_token_count": 654.925, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9908588886260986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0091411042958498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.90465087890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.89415283203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2220, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.19991869777441024, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.017055718787014486, + "grad_norm": 0.13280940055847168, + "kept_token_count": 654.85625, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9908194243907928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009180570393800736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.9734344482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.65187377929686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2230, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.1999979719519615, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.017561471182852985, + "grad_norm": 0.15218600630760193, + "kept_token_count": 654.796875, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9905003607273102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009499643463641405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.6136077880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.179833984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2240, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.20033095479011537, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.01874117348343134, + "grad_norm": 0.14727574586868286, + "kept_token_count": 654.3390625, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9899732947349549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010026718955487013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.5255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.72691345214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.20057826340198517, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.018994948733597992, + "grad_norm": 0.10187821835279465, + "kept_token_count": 654.1203125, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9904416143894196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009558399114757777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.2681671142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.47808532714845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2260, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20043659210205078, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.017461747955530883, + "grad_norm": 0.11183682829141617, + "kept_token_count": 654.1140625, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9904410243034363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009558957815170289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.07291107177736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4398162841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2270, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20038910955190659, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.01542983828112483, + "grad_norm": 0.12572400271892548, + "kept_token_count": 654.284375, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9906045496463776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009395454544574022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.03868103027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.46690368652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2280, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20037826746702195, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.017840153723955154, + "grad_norm": 0.10099869221448898, + "kept_token_count": 654.253125, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9904503405094147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009549649618566036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.2561401367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.5461181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20047180205583573, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.017671169713139534, + "grad_norm": 0.14917117357254028, + "kept_token_count": 653.9625, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9902297139167786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0097702756524086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.2613983154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.92250061035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2300, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.19966927021741868, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.019663078151643275, + "grad_norm": 0.15445251762866974, + "kept_token_count": 655.1359375, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9898736715316773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010126337315887213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.2360092163086, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5446838378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2310, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.2002279669046402, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.018210237752646206, + "grad_norm": 0.15722210705280304, + "kept_token_count": 654.4140625, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9898717999458313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010128198750317096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.18084106445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3699554443359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19954503178596497, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.019966205582022667, + "grad_norm": 0.11728785187005997, + "kept_token_count": 654.9234375, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.990035331249237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009964664001017809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.27455139160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2466094970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2330, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20016713589429855, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.016753742657601833, + "grad_norm": 0.12822052836418152, + "kept_token_count": 654.665625, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9896385073661804, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010361473355442286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.0490417480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.25531005859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19970606118440629, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.02124609798192978, + "grad_norm": 0.10490694642066956, + "kept_token_count": 654.66875, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9899201214313507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010079862549901009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.4194641113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1416259765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2350, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.19932264238595962, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.018804083578288555, + "grad_norm": 0.12366495281457901, + "kept_token_count": 655.1546875, + "learning_rate": 2.993280559640159e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9896104216575623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010389582440257072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.1454315185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.08319091796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20042538344860078, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.01783656533807516, + "grad_norm": 0.12903781235218048, + "kept_token_count": 654.253125, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9898065030574799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010193499270826578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.2596435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8510223388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2370, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2004546821117401, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.01709859408438206, + "grad_norm": 0.1280490905046463, + "kept_token_count": 654.346875, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9912558376789093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00874416008591652, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.71315002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0546478271484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2380, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.1996706485748291, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.018242838233709334, + "grad_norm": 0.13065707683563232, + "kept_token_count": 655.2234375, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9907229602336883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009277021698653697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.1120239257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7578918457031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2390, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.19995299726724625, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.01875518886372447, + "grad_norm": 0.12209779769182205, + "kept_token_count": 654.903125, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9899158656597138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01008413704112172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.90616912841796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.54912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2400, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.20058888345956802, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.016154344007372857, + "grad_norm": 0.12568488717079163, + "kept_token_count": 654.16875, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9905581057071686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00944190789014101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.11155700683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.07947998046876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2410, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2004983901977539, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.022027090284973384, + "grad_norm": 0.1149669662117958, + "kept_token_count": 654.0421875, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0268, + "mgd_cos_sim": 0.9904873907566071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009512588661164046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.69020690917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3893280029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2420, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.20049376040697098, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.018641053326427938, + "grad_norm": 0.16499750316143036, + "kept_token_count": 654.0015625, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9899424612522125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010057538747787476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.06281127929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.82573852539065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2430, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.2007027253508568, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.019274538010358812, + "grad_norm": 0.11570124328136444, + "kept_token_count": 654.20625, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9910085201263428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00899147940799594, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.4361297607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1151428222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2440, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.19976747781038284, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.01917600128799677, + "grad_norm": 0.12514732778072357, + "kept_token_count": 655.0375, + "learning_rate": 2.991800117455e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9899818897247314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010018108226358891, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.71559143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.44451904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2450, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.2004181370139122, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.016545785032212733, + "grad_norm": 0.1326190084218979, + "kept_token_count": 654.6765625, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9899333417415619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010066656675189734, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.41860046386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.8655975341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2460, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.19998513609170915, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.01921884324401617, + "grad_norm": 0.14873111248016357, + "kept_token_count": 655.046875, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.990251362323761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009748642332851887, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.0975402832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2849456787109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2470, + "valid_token_count": 818.7953125 + }, + { + "actual_token_mask_ratio": 0.19909253269433974, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.016362371575087308, + "grad_norm": 0.1009136363863945, + "kept_token_count": 655.38125, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9904447257518768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009555270243436099, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.5550140380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.8125274658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2480, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.20030420571565627, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.02085965657606721, + "grad_norm": 0.1386207789182663, + "kept_token_count": 654.525, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9911484718322754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008851519506424665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.3560791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 357.37485961914064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2490, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.1999383106827736, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.018808124121278524, + "grad_norm": 0.13119253516197205, + "kept_token_count": 654.678125, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9899295210838318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010070480033755302, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.24656677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5260894775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2500, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.19977228939533234, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.018776025623083115, + "grad_norm": 0.11481037735939026, + "kept_token_count": 655.0, + "learning_rate": 2.990731576568372e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9896234095096588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010376592539250851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.1130599975586, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.72645263671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2510, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19958180487155913, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.018529202323406935, + "grad_norm": 0.11541073769330978, + "kept_token_count": 654.8796875, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9903862357139588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009613767359405756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.68504028320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.19411010742186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.19911837428808213, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.01795384017750621, + "grad_norm": 0.10862891376018524, + "kept_token_count": 655.421875, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9906153023242951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009384698700159788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.7292541503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9914611816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2530, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.20078154653310776, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.017719999141991138, + "grad_norm": 0.14841903746128082, + "kept_token_count": 653.834375, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9901238441467285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009876158926635981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.0644989013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.20753784179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2540, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20034104883670806, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.020108288805931806, + "grad_norm": 0.11477161943912506, + "kept_token_count": 654.08125, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9894619047641754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010538105759769678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.99986572265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6388305664062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2550, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.20063632875680923, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.019351178128272296, + "grad_norm": 0.18074774742126465, + "kept_token_count": 654.1109375, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9891753613948822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01082464586943388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.16668701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3944793701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2560, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19882308393716813, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.016755141597241164, + "grad_norm": 0.1690591424703598, + "kept_token_count": 655.734375, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9904382467269898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009561739955097437, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.0695037841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4048645019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2570, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.19971247464418412, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.01917908163741231, + "grad_norm": 0.11734548211097717, + "kept_token_count": 655.071875, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9901205062866211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009879493340849876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.64110412597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1466766357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2580, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.20052440017461776, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.015155666228383779, + "grad_norm": 0.10757152736186981, + "kept_token_count": 654.1109375, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9900444984436035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009955480508506299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.29005126953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.21715393066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2590, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.1996548131108284, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.01998436264693737, + "grad_norm": 0.131780207157135, + "kept_token_count": 654.94375, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9909843981266022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009015592001378536, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.76514587402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6262451171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2600, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19976735711097718, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.018713930062949656, + "grad_norm": 0.12185224145650864, + "kept_token_count": 654.75625, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9902759611606597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009724043682217598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.9852355957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.01026916503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2610, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2002306714653969, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.01866002194583416, + "grad_norm": 0.13857387006282806, + "kept_token_count": 654.1921875, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903168499469757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009683145582675934, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.65558166503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.40684204101564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2620, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.2004443645477295, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.01734399711713195, + "grad_norm": 0.11995842307806015, + "kept_token_count": 654.1984375, + "learning_rate": 2.988398916103648e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9907093167304992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009290700405836105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.18255920410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.09486389160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2630, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19905277490615844, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.01908741006627679, + "grad_norm": 0.148716539144516, + "kept_token_count": 655.6671875, + "learning_rate": 2.988192766402212e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9902076721191406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009792321361601353, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.3529388427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2883270263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2640, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.20044782906770706, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.0191636280156672, + "grad_norm": 0.13697479665279388, + "kept_token_count": 654.15625, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9906365990638732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009363406710326672, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.0712417602539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.63507690429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2003001019358635, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.018708634097129105, + "grad_norm": 0.13556675612926483, + "kept_token_count": 654.28125, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.989925992488861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01007399819791317, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.88458862304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.35333862304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2660, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20009736716747284, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.019042138289660216, + "grad_norm": 0.1549699306488037, + "kept_token_count": 654.3390625, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9898544609546661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010145534295588732, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.4374145507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6076141357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2670, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20016764402389525, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.020329127833247185, + "grad_norm": 0.1548193097114563, + "kept_token_count": 654.440625, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9899658679962158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010034122783690692, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.5984222412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.53937072753905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2680, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2005585789680481, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.019057206716388463, + "grad_norm": 0.13685891032218933, + "kept_token_count": 654.3375, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9904552936553955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009544705413281918, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.42234497070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.38795776367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2690, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20070187300443648, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.019364450685679914, + "grad_norm": 0.13877683877944946, + "kept_token_count": 653.825, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9896354913711548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010364498198032378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.8230041503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.27958984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2700, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.19924246668815612, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.017280679661780596, + "grad_norm": 0.1402895152568817, + "kept_token_count": 655.328125, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9902661621570588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009733855444937944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.1785858154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4010437011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2710, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1994341269135475, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.018105199746787547, + "grad_norm": 0.159075528383255, + "kept_token_count": 654.93125, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9898231089115143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010176873207092286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.8641357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.08600463867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2004133492708206, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.018178317230194807, + "grad_norm": 0.14952297508716583, + "kept_token_count": 654.259375, + "learning_rate": 2.986256076618961e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9903648436069489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00963515480980277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.5542816162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3269989013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2730, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.1995685413479805, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.017085123620927335, + "grad_norm": 0.15318462252616882, + "kept_token_count": 654.7296875, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9904456794261932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009554298035800457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.6312561035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.13527526855466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2740, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.20065852701663972, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.01744743138551712, + "grad_norm": 0.1604301780462265, + "kept_token_count": 653.9203125, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9898355066776275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01016449360176921, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.6587829589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3470825195312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2750, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.19996611773967743, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.016860695835202933, + "grad_norm": 0.13337965309619904, + "kept_token_count": 654.7375, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9905352890491486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009464717842638493, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.65277099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2994903564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2760, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.19886025339365004, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.020708336867392062, + "grad_norm": 0.14571507275104523, + "kept_token_count": 655.253125, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.990506511926651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009493474382907152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 270.04236755371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1617919921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2770, + "valid_token_count": 817.9015625 + }, + { + "actual_token_mask_ratio": 0.1996907353401184, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.01785695543512702, + "grad_norm": 0.15320056676864624, + "kept_token_count": 654.509375, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9903577327728271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009642274025827647, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.99025573730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.06853942871095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2780, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.19908805936574936, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.018027619831264018, + "grad_norm": 0.13915042579174042, + "kept_token_count": 655.2453125, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9897690057754517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010230998322367669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.8738189697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.97431030273435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19918838292360305, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.019264247361570598, + "grad_norm": 0.17068760097026825, + "kept_token_count": 655.0296875, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9908980429172516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009101965371519328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.66810913085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.25494079589845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2800, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.20103665590286254, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.020607704296708107, + "grad_norm": 0.11923182010650635, + "kept_token_count": 653.928125, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9900180280208588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998197365552187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.2912170410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.11919555664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2810, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.20122388154268264, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.019069507252424955, + "grad_norm": 0.12838581204414368, + "kept_token_count": 653.678125, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9898948550224305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010105154942721128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.44365844726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.49794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2820, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19972456097602845, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.016291474737226964, + "grad_norm": 0.11934614926576614, + "kept_token_count": 655.040625, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9905790567398072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009420945309102535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.92900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8454223632813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2830, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.20083381980657578, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.016680398583412172, + "grad_norm": 0.11208625882863998, + "kept_token_count": 653.9625, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9906791985034943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009320820309221744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.23732299804686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.7584716796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2840, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19945904463529587, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.016630162484943867, + "grad_norm": 0.10871332138776779, + "kept_token_count": 655.115625, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9909503519535064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009049654286354781, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.7024444580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4506378173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2850, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19967800974845887, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.02060832241550088, + "grad_norm": 0.12490672618150711, + "kept_token_count": 655.075, + "learning_rate": 2.98320044140705e-05, + "loss": 0.025, + "mgd_cos_sim": 0.991261750459671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873824916779995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.5017028808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4169860839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2860, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.19970024079084397, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.018390125036239623, + "grad_norm": 0.11177120357751846, + "kept_token_count": 654.6859375, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9910174965858459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00898249177262187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.8810241699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6254455566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2870, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20010494589805602, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.01773098427802324, + "grad_norm": 0.10147281736135483, + "kept_token_count": 654.4125, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9902681946754456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009731811843812466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.28282470703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6448486328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2880, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.19989102333784103, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.019734568428248167, + "grad_norm": 0.13470417261123657, + "kept_token_count": 654.703125, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9898108422756196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010189179982990026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.71832580566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8458190917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2890, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19912651777267457, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.017405842896550894, + "grad_norm": 0.11485286056995392, + "kept_token_count": 655.115625, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9905014157295227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009498585667461156, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.97364196777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6749542236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2900, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.2001635894179344, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.01726190112531185, + "grad_norm": 0.1479552984237671, + "kept_token_count": 654.671875, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9904330193996429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009566984605044127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.1133178710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.33741455078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2910, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.19963856935501098, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.020671994797885417, + "grad_norm": 0.13786789774894714, + "kept_token_count": 654.7625, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9891887187957764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010811276827007532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.7742553710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7694030761719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2920, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20023366212844848, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.019058089051395655, + "grad_norm": 0.13288702070713043, + "kept_token_count": 654.7109375, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.024, + "mgd_cos_sim": 0.990196019411087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00980397528037429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.5385375976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.38982849121095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2930, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.200118787586689, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.021584697254002094, + "grad_norm": 0.16292370855808258, + "kept_token_count": 654.5828125, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0264, + "mgd_cos_sim": 0.9903036057949066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009696415439248086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.0968048095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.11619873046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2940, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19980023801326752, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.017404350265860558, + "grad_norm": 0.14279642701148987, + "kept_token_count": 655.025, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9901939272880554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009806086122989655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.51722412109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.15923461914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2950, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.1994458943605423, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.019097087532281877, + "grad_norm": 0.14508895576000214, + "kept_token_count": 655.21875, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9905154705047607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009484517388045788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.98871154785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6507629394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2960, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.1997467577457428, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.017441992461681367, + "grad_norm": 0.12509052455425262, + "kept_token_count": 654.825, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9903774678707122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009622526727616787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.0761993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0194122314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20016099363565446, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.01815900020301342, + "grad_norm": 0.11933688819408417, + "kept_token_count": 654.6, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9893278837203979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010672106686979533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.49998168945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.965185546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19990289062261582, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.016639767959713937, + "grad_norm": 0.1671808362007141, + "kept_token_count": 654.4859375, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9899517476558686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010048246290534735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.12403564453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.35646667480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 2990, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.1999048888683319, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.016614891681820156, + "grad_norm": 0.11216627061367035, + "kept_token_count": 655.040625, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9900536239147186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009946382604539394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.72405700683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3935272216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3000, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.19988944083452226, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.018023655097931624, + "grad_norm": 0.1942165493965149, + "kept_token_count": 654.7984375, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9902044832706451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009795515891164541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.6014709472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.93837280273436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3010, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19973337054252624, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.018681298196315765, + "grad_norm": 0.11017128825187683, + "kept_token_count": 655.2234375, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9888210415840148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.011178962979465723, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.98479919433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6930816650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3020, + "valid_token_count": 818.7578125 + }, + { + "actual_token_mask_ratio": 0.2015963912010193, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.01609750548377633, + "grad_norm": 0.11051293462514877, + "kept_token_count": 653.49375, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9900258243083954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009974176622927189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.5417907714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4377807617187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3030, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.20012594163417816, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.018822292797267437, + "grad_norm": 0.10560665279626846, + "kept_token_count": 654.6828125, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9899300694465637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010069924034178257, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.8280487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.93865966796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3040, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20025965422391892, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.015849865786731244, + "grad_norm": 0.11529863625764847, + "kept_token_count": 654.146875, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9904434025287628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009556600265204907, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.1575958251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3459533691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3050, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.20016442090272904, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.018701671436429024, + "grad_norm": 0.12785276770591736, + "kept_token_count": 654.1921875, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9895651578903198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010434852540493011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.94617614746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2497619628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3060, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.19997950792312622, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.017455340828746557, + "grad_norm": 0.12396464496850967, + "kept_token_count": 654.6375, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9904802799224853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00951971923932433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.54009399414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0973602294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3070, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2002382144331932, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.018145105428993703, + "grad_norm": 0.11781002581119537, + "kept_token_count": 654.2, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9904611587524415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009538830537348986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.72464447021486, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.02342529296874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.2001833587884903, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.018490068148821592, + "grad_norm": 0.15435810387134552, + "kept_token_count": 654.3046875, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912457942962647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008754216134548187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.6194183349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2622039794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3090, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.19897366762161256, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.01975453943014145, + "grad_norm": 0.11445789039134979, + "kept_token_count": 655.2578125, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9900807440280914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009919262770563365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.1602416992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0217590332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3100, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19976753741502762, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.01720416359603405, + "grad_norm": 0.11114414036273956, + "kept_token_count": 654.8828125, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9906261742115021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009373834170401097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.4587371826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9673767089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3110, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20005215853452682, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.017585783917456865, + "grad_norm": 0.11237545311450958, + "kept_token_count": 654.7015625, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9905666470527649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009433335531502962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.99595642089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.5614410400391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3120, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20074853599071502, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.02048236895352602, + "grad_norm": 0.1355680525302887, + "kept_token_count": 653.9, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9904127418994904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009587251767516135, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.51817016601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0546081542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3130, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2001182734966278, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.017629991006106138, + "grad_norm": 0.12426178902387619, + "kept_token_count": 654.51875, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9907402694225311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009259751625359058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.0093994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4864837646484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3140, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20091647952795028, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.01725606843829155, + "grad_norm": 0.10064880549907684, + "kept_token_count": 653.95, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9906553089618683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009344696439802647, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.26109924316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.59714965820314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3150, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.2000082403421402, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.017493548337370157, + "grad_norm": 0.1620635986328125, + "kept_token_count": 654.740625, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9899038910865784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010096108540892601, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.559375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.26934814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3160, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20000490546226501, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.01636023847386241, + "grad_norm": 0.10418173670768738, + "kept_token_count": 654.609375, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9908044576644898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009195542335510254, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.2461303710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0778015136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3170, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20053548067808152, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.016952919960021972, + "grad_norm": 0.10862560570240021, + "kept_token_count": 654.2765625, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9905674576759338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009432550240308046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.4947998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9505859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3180, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20023309886455537, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.018887017294764517, + "grad_norm": 0.1322493702173233, + "kept_token_count": 654.353125, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9899326503276825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010067350044846535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.00091857910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.09217529296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3190, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.20036189556121825, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.01759546808898449, + "grad_norm": 0.0984029471874237, + "kept_token_count": 654.428125, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9901371896266937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009862807113677263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.0745391845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 358.7765319824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3200, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2008933573961258, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.01753950575366616, + "grad_norm": 0.12600407004356384, + "kept_token_count": 653.8671875, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9900117933750152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998820224776864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.35756225585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2664276123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19930297136306763, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.016923049092292787, + "grad_norm": 0.12832264602184296, + "kept_token_count": 655.1109375, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9906112313270569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938876299187541, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.6228393554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5218444824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3220, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.1996975690126419, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.014540248177945614, + "grad_norm": 0.14347514510154724, + "kept_token_count": 654.8234375, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9912081897258759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008791805990040303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.80116577148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4922821044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3230, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2002026453614235, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.017585203889757393, + "grad_norm": 0.13382379710674286, + "kept_token_count": 654.1640625, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9910009622573852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008999035973101855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.212451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1431121826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3240, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.20010227262973784, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.017277567088603972, + "grad_norm": 0.1633240133523941, + "kept_token_count": 654.609375, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9898293793201447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010170617327094077, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.3257110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.30126953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3250, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19967928230762483, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.01893227584660053, + "grad_norm": 0.12994524836540222, + "kept_token_count": 654.840625, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9900813162326813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009918687772005797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.31604919433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.78985900878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3260, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2003158986568451, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.017324628215283154, + "grad_norm": 0.11273682117462158, + "kept_token_count": 654.596875, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9908972442150116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009102756530046463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.7976806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.91184692382814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3270, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1999126449227333, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.0190780533477664, + "grad_norm": 0.12595903873443604, + "kept_token_count": 654.6046875, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9896202206611633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010379779525101185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.79010314941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9059753417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3280, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.19990316331386565, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.022687675058841707, + "grad_norm": 0.13690221309661865, + "kept_token_count": 654.721875, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0278, + "mgd_cos_sim": 0.9897363483905792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010263656917959451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.62579650878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4125946044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3290, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.1993679702281952, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.01798833776265383, + "grad_norm": 0.1428299844264984, + "kept_token_count": 655.2421875, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9905628859996796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009437127131968736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.2408050537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5136291503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3300, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.19973038136959076, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.01781006306409836, + "grad_norm": 0.11313614249229431, + "kept_token_count": 654.803125, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9909850418567657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00901495311409235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.1754150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.966943359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3310, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19945151656866072, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.0202585157006979, + "grad_norm": 0.1666233241558075, + "kept_token_count": 655.140625, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9904579639434814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00954202627763152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.9608184814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.31150207519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3320, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19951759427785873, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.017530067637562752, + "grad_norm": 0.1556137353181839, + "kept_token_count": 654.959375, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9905709385871887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009429062716662884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.3388946533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4237365722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3330, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.1992616280913353, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.01870952993631363, + "grad_norm": 0.1141478419303894, + "kept_token_count": 654.96875, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9905878841876984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009412114974111319, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.09296264648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1761871337891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3340, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.20056968927383423, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.01741895666345954, + "grad_norm": 0.13125254213809967, + "kept_token_count": 653.9703125, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9903712093830108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009628791734576226, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.9716857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4803070068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3350, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.20023189038038253, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.01759707583114505, + "grad_norm": 0.12298611551523209, + "kept_token_count": 654.2453125, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9900736093521119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009926379192620516, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.2957336425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.94965209960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3360, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.19957306683063508, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.019066819362342356, + "grad_norm": 0.1276680827140808, + "kept_token_count": 654.9171875, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9901982307434082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009801786486059427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.05645141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3027587890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3370, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.20055125802755355, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.015188505500555038, + "grad_norm": 0.1168186217546463, + "kept_token_count": 654.26875, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9905723690986633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009427625872194766, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.9052032470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.8838775634766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3380, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20029156357049943, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.01579436715692282, + "grad_norm": 0.12398607283830643, + "kept_token_count": 654.26875, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9898600041866302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010139993391931057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.67706909179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5563934326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3390, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.19968430548906327, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.01990048252046108, + "grad_norm": 0.14715513586997986, + "kept_token_count": 654.5875, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9900877594947814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00991224693134427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.0317687988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9765258789063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3400, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.1995129704475403, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.015591507125645877, + "grad_norm": 0.13071244955062866, + "kept_token_count": 654.903125, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9906735062599182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009326503239572048, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.78941345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.81455383300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3410, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19942530244588852, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.015307436976581813, + "grad_norm": 0.14805306494235992, + "kept_token_count": 655.0640625, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9898462772369385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010153718013316392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.99879455566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3625213623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3420, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.19916752576828003, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.020286644343286753, + "grad_norm": 0.12935447692871094, + "kept_token_count": 655.1859375, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9899767398834228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010023253038525581, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.9544281005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0545989990234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3430, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19919439107179643, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.021116176433861256, + "grad_norm": 0.1076912134885788, + "kept_token_count": 654.990625, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.989626544713974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010373466927558184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.6787994384766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0919677734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3440, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.2001492291688919, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.018646615277975797, + "grad_norm": 0.12281975895166397, + "kept_token_count": 654.7625, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.989883291721344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01011670958250761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.7428436279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4746795654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3450, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.1996383935213089, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.019741767644882204, + "grad_norm": 0.1384584903717041, + "kept_token_count": 654.59375, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9907229900360107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00927701499313116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.27520141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.70142517089846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3460, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.200251105427742, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.016687629278749228, + "grad_norm": 0.10905281454324722, + "kept_token_count": 654.5515625, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9899162769317627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010083729773759842, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.6611053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.60130615234374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3470, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19971101433038713, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.019050707388669253, + "grad_norm": 0.15428142249584198, + "kept_token_count": 655.1625, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9907315909862519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009268415253609419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.05870056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6116516113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3480, + "valid_token_count": 818.6578125 + }, + { + "actual_token_mask_ratio": 0.19934158772230148, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.01673695147037506, + "grad_norm": 0.13453125953674316, + "kept_token_count": 655.1703125, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9906914830207825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009308507107198239, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.73212280273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.76881408691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3490, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.2002188578248024, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.017223554011434317, + "grad_norm": 0.15119047462940216, + "kept_token_count": 654.515625, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9903013825416564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00969862649217248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.28522338867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0985809326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3500, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.19972306191921235, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.017040603049099445, + "grad_norm": 0.10457143932580948, + "kept_token_count": 654.7796875, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9906523525714874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009347642213106156, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.0209197998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3638641357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3510, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.201168093085289, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.020043176505714654, + "grad_norm": 0.13928958773612976, + "kept_token_count": 653.71875, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9909081399440766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009091864246875048, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.4970764160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.06947021484376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3520, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.199602909386158, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.019314936269074677, + "grad_norm": 0.1263510286808014, + "kept_token_count": 655.040625, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9907346487045288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009265354834496975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.31434326171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7279815673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2004316046833992, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01754040624946356, + "grad_norm": 0.12532645463943481, + "kept_token_count": 654.63125, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9902559041976928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009744109399616719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.2064666748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10455627441405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3540, + "valid_token_count": 818.73125 + }, + { + "actual_token_mask_ratio": 0.19982946366071702, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.017866384610533713, + "grad_norm": 0.12403100728988647, + "kept_token_count": 654.734375, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9896686017513275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010331392288208008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.7854339599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.40771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3550, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20059781074523925, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.018323629349470138, + "grad_norm": 0.19160665571689606, + "kept_token_count": 654.034375, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9902245223522186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009775473829358816, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6844818115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.848193359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3560, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20044585168361664, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.018566618021577597, + "grad_norm": 0.15372824668884277, + "kept_token_count": 654.190625, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9905782103538513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009421789739280938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.89207763671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5664489746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2006010591983795, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.019643317349255084, + "grad_norm": 0.114413321018219, + "kept_token_count": 654.3265625, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9901354730129241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009864514600485564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.4485168457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2502746582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3580, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.19999477118253708, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.017697517201304434, + "grad_norm": 0.11325004696846008, + "kept_token_count": 654.70625, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.990256541967392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009743464365601539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.5719909667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3016052246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3590, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.20041269361972808, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.01963293682783842, + "grad_norm": 0.13818052411079407, + "kept_token_count": 654.3109375, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9900335431098938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009966454096138477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.48285217285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.5773498535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3600, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.1996481880545616, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.017018522694706918, + "grad_norm": 0.1360412836074829, + "kept_token_count": 654.696875, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9905309438705444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00946905668824911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.2626678466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.94976501464845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3610, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20059642642736436, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.01572856092825532, + "grad_norm": 0.11962161958217621, + "kept_token_count": 653.7734375, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9907244622707367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009275534749031067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.54579467773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.818017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3620, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.19962747097015382, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.016739476285874844, + "grad_norm": 0.16192080080509186, + "kept_token_count": 654.765625, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9904488861560822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009551114868372679, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.70159912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4091339111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3630, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.20022935420274734, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.017753967456519602, + "grad_norm": 0.12794461846351624, + "kept_token_count": 654.6703125, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9899389803409576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010061030089855195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.9233917236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3674774169922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3640, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.20034151375293732, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.019665971770882605, + "grad_norm": 0.19345498085021973, + "kept_token_count": 654.26875, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.98996142745018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010038556531071663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.57216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3597076416016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3650, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.20046194344758989, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.018534077610820533, + "grad_norm": 0.17755720019340515, + "kept_token_count": 654.24375, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9903749942779541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00962501298636198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 281.995703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7968017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20033816993236542, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.018970733880996703, + "grad_norm": 0.1396554559469223, + "kept_token_count": 654.3640625, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9900751948356629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00992480842396617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.57070007324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2991424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3670, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20022678524255752, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.016228895634412766, + "grad_norm": 0.13774798810482025, + "kept_token_count": 654.725, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9895242393016815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010475753992795944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.7958099365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3680, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.20016996711492538, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.01747508505359292, + "grad_norm": 0.12303568422794342, + "kept_token_count": 654.4609375, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9903990149497985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009600988123565912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.59070739746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1829376220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3690, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2002018854022026, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.018904855009168386, + "grad_norm": 0.1090935543179512, + "kept_token_count": 654.7265625, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9904650390148163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009534980915486813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.1050720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.5415313720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3700, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.19946447461843492, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.018615889828652143, + "grad_norm": 0.1144830510020256, + "kept_token_count": 654.98125, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9896534562110901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010346552357077599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.2263427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.75501708984376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3710, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20054257661104202, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.019510726630687713, + "grad_norm": 0.10808004438877106, + "kept_token_count": 654.2734375, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.989802622795105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010197380743920802, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.1332672119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1393981933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3720, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2007169783115387, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.01809632396325469, + "grad_norm": 0.10492069274187088, + "kept_token_count": 654.0171875, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9898750960826874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01012489916756749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.7805633544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.99049377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3730, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.1991270437836647, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.0175626871176064, + "grad_norm": 0.10272404551506042, + "kept_token_count": 655.3796875, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9901351630687714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009864835347980262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.5284820556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.00156860351564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3740, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.1995532050728798, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.01920426730066538, + "grad_norm": 0.12728233635425568, + "kept_token_count": 654.6421875, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9905233979225159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009476595651358366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.70140075683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2891326904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3750, + "valid_token_count": 817.840625 + }, + { + "actual_token_mask_ratio": 0.20034815818071366, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.018104203417897223, + "grad_norm": 0.13846895098686218, + "kept_token_count": 654.3265625, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9902371108531952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00976290162652731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.9255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2127349853516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3760, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19977978616952896, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.017322477605193853, + "grad_norm": 0.10918847471475601, + "kept_token_count": 654.596875, + "learning_rate": 2.953325994871975e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9906816184520721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009318382572382688, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.447607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.66710205078124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3770, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19991613030433655, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.01646899925544858, + "grad_norm": 0.14058493077754974, + "kept_token_count": 654.7953125, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.990423047542572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009576955158263446, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.4726226806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1923828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3780, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.1987566828727722, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.018171268049627542, + "grad_norm": 0.12859982252120972, + "kept_token_count": 655.5203125, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906785070896149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0093214999884367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.7373077392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.20415649414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3790, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.1992592841386795, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.018650664947926998, + "grad_norm": 0.1586649864912033, + "kept_token_count": 655.1328125, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9903806507587433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009619351755827666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.2223266601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.41785583496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3800, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2010032504796982, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.016280369926244022, + "grad_norm": 0.12436715513467789, + "kept_token_count": 653.59375, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9903392672538758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009660723060369492, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.22057189941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.75384216308595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3810, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19938739091157914, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.018348973523825408, + "grad_norm": 0.11243964731693268, + "kept_token_count": 655.11875, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9905312538146973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009468738548457622, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.3290496826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.62359313964845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3820, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20079096853733064, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.018323609326034784, + "grad_norm": 0.10838554799556732, + "kept_token_count": 654.1125, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908343017101288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009165703784674406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.3106719970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3258941650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3830, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.20071546584367753, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.01746600866317749, + "grad_norm": 0.10463828593492508, + "kept_token_count": 653.9890625, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.990590512752533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0094094792380929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.7554016113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4021911621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3840, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.19964669942855834, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.017370970733463765, + "grad_norm": 0.11806047707796097, + "kept_token_count": 654.9734375, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9904086232185364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009591359831392764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.9730224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.57342529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3850, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20037002116441727, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.01787464814260602, + "grad_norm": 0.13633950054645538, + "kept_token_count": 654.0953125, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9899155020713806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010084502585232257, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.5750701904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.97690734863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3860, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20046992897987365, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.019687449652701618, + "grad_norm": 0.13315150141716003, + "kept_token_count": 654.228125, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9904785454273224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009521457459777593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.68617248535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9326110839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3870, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19929812252521514, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.018055344745516778, + "grad_norm": 0.12375845015048981, + "kept_token_count": 655.1328125, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9909878075122833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0090121959336102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.2399627685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0843780517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3880, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.20021283030509948, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.01695128148421645, + "grad_norm": 0.13665471971035004, + "kept_token_count": 654.44375, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9904050827026367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009594917390495538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.82118225097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6998992919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3890, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20007189214229584, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.018608549423515797, + "grad_norm": 0.13739599287509918, + "kept_token_count": 654.70625, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9897834420204162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010216562263667583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.89051513671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4602355957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3900, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20003060400485992, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.020553353987634183, + "grad_norm": 0.12921597063541412, + "kept_token_count": 654.6984375, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.990356731414795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009643268957734107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.95224914550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2384979248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3910, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.2005121648311615, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.01974395280703902, + "grad_norm": 0.14305078983306885, + "kept_token_count": 654.190625, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9907638847827911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009236111305654049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 322.7481658935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.05519104003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3920, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.1994104340672493, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.016388456244021653, + "grad_norm": 0.1328289657831192, + "kept_token_count": 654.6875, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9911906480789184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008809350989758968, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.5350830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4435119628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3930, + "valid_token_count": 817.759375 + }, + { + "actual_token_mask_ratio": 0.20030089616775512, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.016971901711076497, + "grad_norm": 0.13962829113006592, + "kept_token_count": 654.275, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9902270495891571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009772948827594519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.92071838378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0669799804688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20058328211307525, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.016915835626423358, + "grad_norm": 0.14225749671459198, + "kept_token_count": 654.0890625, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9907231628894806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009276840183883906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.2411224365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.551953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3950, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19981342703104019, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.0191386915743351, + "grad_norm": 0.1347896307706833, + "kept_token_count": 654.6671875, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9896517872810364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01034820070490241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.3008972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3941925048828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3960, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19997411966323853, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.017860427126288415, + "grad_norm": 0.13028189539909363, + "kept_token_count": 654.640625, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9901800811290741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00981990797445178, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.1270050048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7700561523437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.1992184042930603, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.023857644014060497, + "grad_norm": 0.14524385333061218, + "kept_token_count": 655.365625, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0285, + "mgd_cos_sim": 0.9906319558620453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0093680365011096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.1747619628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.17376403808595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3980, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.20065232515335082, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.021392638515681028, + "grad_norm": 0.16153304278850555, + "kept_token_count": 654.0078125, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9901776850223541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009822309762239457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.2787628173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9962463378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 3990, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19909581691026687, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.01851179776713252, + "grad_norm": 0.0991782397031784, + "kept_token_count": 655.0390625, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9907828390598297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009217162244021892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.9766082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1210479736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4000, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.20073545575141907, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.019760741107165813, + "grad_norm": 0.1468106061220169, + "kept_token_count": 653.9, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9904357731342316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009564224071800709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.5069213867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.79344482421874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4010, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.1996154770255089, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.017227212619036435, + "grad_norm": 0.1400209218263626, + "kept_token_count": 655.1625, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9910276770591736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008972315303981303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.0652191162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5757690429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4020, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.20019519180059434, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.01631419900804758, + "grad_norm": 0.12431526184082031, + "kept_token_count": 654.625, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9906721889972687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009327801689505578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.51416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2236846923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4030, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.1987767443060875, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.017790063749998808, + "grad_norm": 0.12829895317554474, + "kept_token_count": 655.4078125, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9901278734207153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009872120805084705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.1028778076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1346954345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4040, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.20060869455337524, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.01597069324925542, + "grad_norm": 0.1276606172323227, + "kept_token_count": 654.234375, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9907553434371948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009244666155427694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.8364318847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.490771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19942981153726577, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.017537553049623966, + "grad_norm": 0.12318260967731476, + "kept_token_count": 654.9375, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907471358776092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009252853412181138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.9004302978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6385192871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4060, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20013499706983567, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.017850416339933874, + "grad_norm": 0.12003554403781891, + "kept_token_count": 654.49375, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.990379923582077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009620072040706872, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.448291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.44801940917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4070, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19997323751449586, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.01784026138484478, + "grad_norm": 0.11444376409053802, + "kept_token_count": 654.8078125, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.990300452709198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009699530526995659, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.02452087402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.434765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4080, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.1992183580994606, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.017646242305636405, + "grad_norm": 0.1744135469198227, + "kept_token_count": 654.928125, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9902702152729035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009729775693267584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.2687652587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.29392395019534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4090, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.20100397020578384, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.018529738020151853, + "grad_norm": 0.12416407465934753, + "kept_token_count": 653.734375, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9900517225265503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009948276355862618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.9039245605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.87091064453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4100, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2011294409632683, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.01822133418172598, + "grad_norm": 0.10817138850688934, + "kept_token_count": 654.03125, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9900804936885834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009919514413923025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.17853088378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.309326171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4110, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.2007441744208336, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.01759825125336647, + "grad_norm": 0.1299762725830078, + "kept_token_count": 653.93125, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9904402792453766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009559718798846007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.61670837402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.79408569335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4120, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19960351586341857, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.015444173756986857, + "grad_norm": 0.11605347692966461, + "kept_token_count": 654.890625, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9906870782375335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009312915708869695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.8386260986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6349761962891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4130, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20012758523225785, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.02174734426662326, + "grad_norm": 0.11117541790008545, + "kept_token_count": 654.5921875, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9904371857643127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009562809299677611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6441986083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.32373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4140, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.1996051177382469, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.018214864190667868, + "grad_norm": 0.1330231875181198, + "kept_token_count": 654.7984375, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9895436406135559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010456349514424802, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.6179626464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.93580017089846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4150, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.19972628057003022, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.019864359125494956, + "grad_norm": 0.11686307191848755, + "kept_token_count": 654.7875, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9898541212081909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010145862121134996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.05823669433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7051635742188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4160, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2001463770866394, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.018487141560763122, + "grad_norm": 0.11797965317964554, + "kept_token_count": 654.28125, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.990060567855835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009939435869455338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.4708618164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.01636657714846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4170, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.20025981813669205, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.01773261735215783, + "grad_norm": 0.12578897178173065, + "kept_token_count": 654.3640625, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9902453422546387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009754668455570936, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 307.89641723632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.92027587890624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4180, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2012068212032318, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.016583999246358873, + "grad_norm": 0.10214100033044815, + "kept_token_count": 653.8640625, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9899070858955383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010092906095087529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.1989410400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3644226074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4190, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.20052131712436677, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.016240899730473757, + "grad_norm": 0.1397363245487213, + "kept_token_count": 654.09375, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9898428499698639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01015715254470706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.38519592285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8280395507812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4200, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.19943805783987045, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.017118064872920512, + "grad_norm": 0.11644476652145386, + "kept_token_count": 655.0125, + "learning_rate": 2.933616574579118e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9902657091617584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009734295401722192, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.9309783935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.93519287109376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4210, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2003009185194969, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.016302976105362178, + "grad_norm": 0.12907293438911438, + "kept_token_count": 654.5375, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9893873393535614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010612662695348263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.1303985595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4779388427734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4220, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.19957195818424225, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.016671433672308923, + "grad_norm": 0.13683398067951202, + "kept_token_count": 655.2703125, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9895985305309296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010401470214128494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.946240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.87470092773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4230, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.19925273209810257, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.018888262473046778, + "grad_norm": 0.14202411472797394, + "kept_token_count": 655.2875, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9909201443195343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009079858660697937, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.7733489990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3855682373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4240, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2005894348025322, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.016801489051431417, + "grad_norm": 0.11654939502477646, + "kept_token_count": 654.14375, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9907371282577515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009262877702713012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.4311187744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.42308349609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4250, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19984342753887177, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.016772313602268697, + "grad_norm": 0.12431009113788605, + "kept_token_count": 654.753125, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9905544102191925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009445579908788203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.5995635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.59181518554686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4260, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.1999206453561783, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.014818094018846751, + "grad_norm": 0.12038257718086243, + "kept_token_count": 654.615625, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9905095815658569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009490406233817339, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.9752593994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.00966186523436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4270, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.199143685400486, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.01845128843560815, + "grad_norm": 0.16638998687267303, + "kept_token_count": 655.365625, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9899952292442322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010004778951406479, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.74151000976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.7090270996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4280, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20036682933568956, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.019018673803657293, + "grad_norm": 0.1789345145225525, + "kept_token_count": 654.39375, + "learning_rate": 2.929669297182963e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9899640023708344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010035993997007608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.94102172851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0558197021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4290, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19988016784191132, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.01817280724644661, + "grad_norm": 0.11793210357427597, + "kept_token_count": 654.6046875, + "learning_rate": 2.929168063845373e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9903404235839843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009659580420702696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.9007934570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.17047119140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4300, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19922105967998505, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.018415904976427554, + "grad_norm": 0.13796602189540863, + "kept_token_count": 655.4515625, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9891587436199188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010841267183423042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.2961090087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7293731689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4310, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2009296953678131, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.016342795453965664, + "grad_norm": 0.1048591360449791, + "kept_token_count": 653.76875, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9901705503463745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00982944667339325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.5752380371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6046630859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4320, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2006787806749344, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.018712608143687248, + "grad_norm": 0.14190879464149475, + "kept_token_count": 654.0234375, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9898623645305633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010137621592730284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.0806884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8678833007813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4330, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20088331699371337, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.019606798700988293, + "grad_norm": 0.14513042569160461, + "kept_token_count": 653.5390625, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.989577180147171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010422820132225752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.1048889160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5280426025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4340, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.20071514397859574, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.01726924246177077, + "grad_norm": 0.10731140524148941, + "kept_token_count": 653.96875, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9901214838027954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009878515359014272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.44237365722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.14284362792966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19930665642023088, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.017472990602254868, + "grad_norm": 0.1338886320590973, + "kept_token_count": 655.015625, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9901531159877777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009846885409206151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.1197113037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.50035705566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4360, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20056209117174148, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.01768427425995469, + "grad_norm": 0.1486426591873169, + "kept_token_count": 654.1984375, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9909591197967529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040893986821175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.3268096923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3759735107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4370, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.20027542412281035, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.019683759938925505, + "grad_norm": 0.14492610096931458, + "kept_token_count": 654.353125, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9896914958953857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010308506619185209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.4921478271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6747802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4380, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2003397062420845, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.018549527041614056, + "grad_norm": 0.1209442988038063, + "kept_token_count": 654.5265625, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9904511630535126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009548832196742296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.53786010742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.40040893554686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4390, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.1991957738995552, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.019182118866592646, + "grad_norm": 0.1426348239183426, + "kept_token_count": 655.5125, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9902041137218476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009795877337455749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.05881958007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3992645263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4400, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.19993885308504106, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.020163962710648776, + "grad_norm": 0.16249556839466095, + "kept_token_count": 654.91875, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9901880562305451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009811940882354975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.05971374511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.92870178222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.20050079822540284, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.018458850495517255, + "grad_norm": 0.11206797510385513, + "kept_token_count": 654.5125, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9904811561107636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009518843702971935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.7375152587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.63807983398436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4420, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.200035984814167, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.019189259316772223, + "grad_norm": 0.16711774468421936, + "kept_token_count": 654.584375, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9898409366607666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010159073956310749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.5321075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.14708251953124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4430, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2001731738448143, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.018453419487923384, + "grad_norm": 0.11241047829389572, + "kept_token_count": 654.7, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9897109806537628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010289025120437145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.7706787109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.19293212890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4440, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.19932434260845183, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.01777347978204489, + "grad_norm": 0.125180184841156, + "kept_token_count": 655.1359375, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9902321517467498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009767845831811428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.4443725585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.13669738769534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.20016997158527375, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.01605059104040265, + "grad_norm": 0.12027190625667572, + "kept_token_count": 654.1140625, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9905462086200714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009453799109905958, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.2333526611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.15204467773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4460, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.20021694302558898, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.016311001498252153, + "grad_norm": 0.11733514070510864, + "kept_token_count": 654.3828125, + "learning_rate": 2.92038190644196e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9905831933021545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009416794963181018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.4557281494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3894348144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4470, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.1997729644179344, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.019183756411075593, + "grad_norm": 0.15718261897563934, + "kept_token_count": 654.790625, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.990056324005127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009943668823689222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.1916198730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8248321533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4480, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19952720999717713, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.019078000355511902, + "grad_norm": 0.15263807773590088, + "kept_token_count": 654.8203125, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.989408153295517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01059184167534113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.5451416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9130035400391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4490, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.2004696950316429, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.019638982135802507, + "grad_norm": 0.16195710003376007, + "kept_token_count": 654.0953125, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.990524685382843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009475296083837747, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.1322967529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4122619628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4500, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19985191822052, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.015090054832398891, + "grad_norm": 0.12340400367975235, + "kept_token_count": 654.81875, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9899046778678894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010095313657075167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.98497314453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.8842803955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4510, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20018871873617172, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.016464709863066675, + "grad_norm": 0.12084690481424332, + "kept_token_count": 654.378125, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9900122046470642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998779507353902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.3330017089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9247711181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4520, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19985462427139283, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.016846766602247952, + "grad_norm": 0.12757082283496857, + "kept_token_count": 654.534375, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9898941814899445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010105821955949067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.19404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.76450805664064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.19982991069555284, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.017834772076457738, + "grad_norm": 0.12979809939861298, + "kept_token_count": 654.7046875, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9893580138683319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010641984455287457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.5761474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.07008361816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4540, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2005698025226593, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.01857809107750654, + "grad_norm": 0.1308911293745041, + "kept_token_count": 654.1671875, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9899678647518158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010032134223729372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 306.38653564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6325439453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4550, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2003049686551094, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.01862712763249874, + "grad_norm": 0.13589468598365784, + "kept_token_count": 654.2625, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9904330849647522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009566913638263942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.9588928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.46034851074216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4560, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.19959096759557723, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.016122028417885303, + "grad_norm": 0.12086667120456696, + "kept_token_count": 654.928125, + "learning_rate": 2.914980384035362e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9902652978897095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009734699502587318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.55885009765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.98607482910154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4570, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19897413849830628, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.01972388681024313, + "grad_norm": 0.10968624800443649, + "kept_token_count": 655.1921875, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9905245006084442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009475491661578416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.79344177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8665771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4580, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.2000620409846306, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.020611412171274424, + "grad_norm": 0.1379229873418808, + "kept_token_count": 654.6625, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9900290727615356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00997094577178359, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.49144287109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.451416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4590, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.1994202435016632, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.016194885410368443, + "grad_norm": 0.12694524228572845, + "kept_token_count": 654.871875, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9907759308815003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009224057942628861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.34722900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.77601318359376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4600, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.1998420014977455, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.019779072795063257, + "grad_norm": 0.11514228582382202, + "kept_token_count": 654.71875, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9901204407215118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009879557322710753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.0006805419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0039794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4610, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20008265674114228, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.01751907095313072, + "grad_norm": 0.1369384527206421, + "kept_token_count": 654.3875, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9902487874031067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009751188196241855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.809619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5193206787109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4620, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.2002474308013916, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.017222224734723568, + "grad_norm": 0.11352996528148651, + "kept_token_count": 654.5109375, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9905821204185485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009417883306741714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.7551300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.67481689453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4630, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19958478063344956, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.015822412725538016, + "grad_norm": 0.10247912257909775, + "kept_token_count": 655.025, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9907421290874481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009257884416729211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.7318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6796447753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4640, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20047754049301147, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.017436011973768473, + "grad_norm": 0.11324824392795563, + "kept_token_count": 654.134375, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9903976500034333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009602353349328041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.9820495605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.15350036621095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4650, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19996324479579924, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.01665884191170335, + "grad_norm": 0.13119862973690033, + "kept_token_count": 654.8328125, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9909510254859925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009048988111317158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.56928100585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1459075927734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4660, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.19997706413269042, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.018481441028416156, + "grad_norm": 0.12157589197158813, + "kept_token_count": 654.575, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9915360629558563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008463929686695338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.40211486816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2270263671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4670, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2001558631658554, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.016914431098848583, + "grad_norm": 0.16040831804275513, + "kept_token_count": 654.628125, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9910172939300537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008982699085026979, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.82883911132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2351898193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4680, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20023613423109055, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.01917656688019633, + "grad_norm": 0.15082918107509613, + "kept_token_count": 654.459375, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9907126903533936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009287315141409635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.6398956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.74175720214845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4690, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19968047738075256, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.019541389867663385, + "grad_norm": 0.11924232542514801, + "kept_token_count": 654.7390625, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9902405500411987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009759458713233471, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.05739440917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.787646484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4700, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20061642229557036, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.018314868304878475, + "grad_norm": 0.16341927647590637, + "kept_token_count": 654.1484375, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9897719621658325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010228040814399719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.94866638183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.89869384765626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4710, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.1995732605457306, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.018618814926594496, + "grad_norm": 0.1440807431936264, + "kept_token_count": 655.3484375, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9896781325340271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010321872122585774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8147338867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.82086791992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4720, + "valid_token_count": 818.75 + }, + { + "actual_token_mask_ratio": 0.19963430017232894, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.018339287023991346, + "grad_norm": 0.14637210965156555, + "kept_token_count": 654.7234375, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9911846697330475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008815344702452422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.0105041503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.59052124023435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4730, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.20011019110679626, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.017694938555359842, + "grad_norm": 0.09026996791362762, + "kept_token_count": 654.4953125, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9905082762241364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009491726476699113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.7853179931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7288818359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4740, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.19975543171167373, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.020302102062851192, + "grad_norm": 0.17021657526493073, + "kept_token_count": 654.9875, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9899900436401368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010009953007102012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.05751037597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.92759094238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4750, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.2004733845591545, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.01795429699122906, + "grad_norm": 0.13527244329452515, + "kept_token_count": 654.090625, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9907907009124756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009209308959543704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.89845581054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.58827209472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.1992231488227844, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.01856919890269637, + "grad_norm": 0.14231985807418823, + "kept_token_count": 655.1046875, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9901368856430054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009863117802888155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.9244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8456756591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4770, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19990838766098024, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.020058566797524692, + "grad_norm": 0.12137661874294281, + "kept_token_count": 654.7515625, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9901510834693908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009848922304809094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.52223205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.58885498046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4780, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20059512555599213, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.019925905112177133, + "grad_norm": 0.131123885512352, + "kept_token_count": 654.20625, + "learning_rate": 2.902492790024129e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9897970020771026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010203013196587563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.3616455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4462585449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4790, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20001301616430284, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.015726257301867008, + "grad_norm": 0.1430777609348297, + "kept_token_count": 654.8890625, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9911726295948029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0088273742236197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.90484008789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.388232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4800, + "valid_token_count": 818.621875 + }, + { + "actual_token_mask_ratio": 0.19943858087062835, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.0202534606680274, + "grad_norm": 0.14915724098682404, + "kept_token_count": 655.1921875, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9908728837966919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009127115551382303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.0708923339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5577453613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4810, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20076896995306015, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.018264339305460453, + "grad_norm": 0.1361066848039627, + "kept_token_count": 654.0375, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9906079649925232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009392036776989698, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.18495178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8774169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4820, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2003461480140686, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.018945355992764236, + "grad_norm": 0.1242927685379982, + "kept_token_count": 654.49375, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9902193605899811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009780640155076981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.27406616210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0920867919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4830, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.20059038549661637, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.01920086545869708, + "grad_norm": 0.11334531754255295, + "kept_token_count": 654.15625, + "learning_rate": 2.899539401701046e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9903479814529419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009652023017406464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.36985168457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6026184082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4840, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19983914792537688, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.01697263913229108, + "grad_norm": 0.11124464869499207, + "kept_token_count": 654.853125, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.990951019525528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009048985503613948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.9778747558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8697570800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4850, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.200176402926445, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.01916957451030612, + "grad_norm": 0.11280794441699982, + "kept_token_count": 654.5734375, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9904979825019836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009502028487622738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.0416320800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.93813171386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4860, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20067697614431382, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.01888089943677187, + "grad_norm": 0.10278332978487015, + "kept_token_count": 654.0828125, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9905813038349152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009418700356036425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.71119995117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.59364013671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4870, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20027577877044678, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.01773131275549531, + "grad_norm": 0.1052008718252182, + "kept_token_count": 654.2625, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9905594706535339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009440520964562892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.0619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.93164978027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4880, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19965635985136032, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.016933836601674558, + "grad_norm": 0.09632936120033264, + "kept_token_count": 654.803125, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9905703365802765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009429662581533193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.79242248535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6787628173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4890, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19979506433010102, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.016882900893688203, + "grad_norm": 0.09576228260993958, + "kept_token_count": 654.8875, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9905602633953094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009439733158797026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.75307312011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7924560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4900, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.200669127702713, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.016453554201871157, + "grad_norm": 0.1218576580286026, + "kept_token_count": 653.8703125, + "learning_rate": 2.895333254526303e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9909148395061493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009085161052644253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.990771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4283874511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4910, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.1995146930217743, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.021538053546100853, + "grad_norm": 0.12156498432159424, + "kept_token_count": 654.7859375, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0267, + "mgd_cos_sim": 0.989635169506073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010364834126085043, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9088562011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0454833984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4920, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.2004389464855194, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.01959020718932152, + "grad_norm": 0.13224481046199799, + "kept_token_count": 654.3140625, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9913738012313843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008626197651028632, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.2357452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8632781982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4930, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.2007785439491272, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.01837355997413397, + "grad_norm": 0.17196230590343475, + "kept_token_count": 654.128125, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9898064017295838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010193612519651651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.5630737304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1468994140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4940, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.20026357769966124, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.0178374994546175, + "grad_norm": 0.10658538341522217, + "kept_token_count": 654.36875, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9908706486225128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009129353798925877, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.4870666503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6451843261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.19961802065372466, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.01731612216681242, + "grad_norm": 0.112886942923069, + "kept_token_count": 655.1078125, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9903260350227356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009673971869051456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.5563659667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1939208984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4960, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.19964714348316193, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.018078806437551977, + "grad_norm": 0.1113860234618187, + "kept_token_count": 654.8859375, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9905978560447692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009402142371982336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.15968627929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.82972717285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4970, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.19954857975244522, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.01843832042068243, + "grad_norm": 0.1257537305355072, + "kept_token_count": 654.8953125, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9910301268100739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008969876822084189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.8550567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8559234619141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20085736662149428, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.018213558848947285, + "grad_norm": 0.13624386489391327, + "kept_token_count": 653.8390625, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9909401357173919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009059868194162846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.68397827148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.96227111816404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 4990, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20030796229839326, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.0182921607978642, + "grad_norm": 0.10412447899580002, + "kept_token_count": 654.5953125, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9906805217266083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009319488238543273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.93610229492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7080810546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5000, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.20097489058971404, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.018977954145520927, + "grad_norm": 0.1292593628168106, + "kept_token_count": 653.9484375, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9908292949199676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009170710574835539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.75990600585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9930023193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5010, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19990647882223128, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.019303036108613014, + "grad_norm": 0.18043647706508636, + "kept_token_count": 654.6796875, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9906562745571137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009343733079731464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.8942565917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0583953857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20080183297395707, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.01844686483964324, + "grad_norm": 0.1513238549232483, + "kept_token_count": 654.1984375, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9908052682876587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00919471736997366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.1123962402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.30963134765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5030, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20111794769763947, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.01761842370033264, + "grad_norm": 0.11225612461566925, + "kept_token_count": 653.8546875, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908112704753875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009188725799322128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.2294006347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4344512939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5040, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.19995283335447311, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.01682629734277725, + "grad_norm": 0.13422542810440063, + "kept_token_count": 654.9234375, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9905711889266968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00942881554365158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.0046020507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9407196044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5050, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.19957887828350068, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.016487256437540055, + "grad_norm": 0.16435624659061432, + "kept_token_count": 654.7984375, + "learning_rate": 2.886040660049283e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9910020947456359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008997914846986532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.94487915039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8177764892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5060, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.19965831190347672, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.016331506706774235, + "grad_norm": 0.12847138941287994, + "kept_token_count": 655.0703125, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9908955335617066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009104481432586908, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.30364685058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.15782165527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5070, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19994204193353654, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.01705649858340621, + "grad_norm": 0.11550905555486679, + "kept_token_count": 654.8203125, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9903878152370453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009612178523093462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.90386352539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.99647827148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.1994314596056938, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.017956823483109473, + "grad_norm": 0.13566648960113525, + "kept_token_count": 655.0234375, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9909167230129242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009083273727446795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.55177001953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.530322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5090, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19975255727767943, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.017328968923538923, + "grad_norm": 0.12019047141075134, + "kept_token_count": 654.6125, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9904405891895294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009559419471770525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.3534301757812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9719940185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.20007603466510773, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.019357914850115777, + "grad_norm": 0.13865910470485687, + "kept_token_count": 654.55625, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9902650356292725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009734975546598435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.0990753173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8443878173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19917465299367904, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.017876662872731687, + "grad_norm": 0.14533618092536926, + "kept_token_count": 655.20625, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9896486103534698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010351384710520505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.24843139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7478759765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5120, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.199464912712574, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.018144538439810277, + "grad_norm": 0.1382836550474167, + "kept_token_count": 655.2140625, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906799793243408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009320020023733378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.31591796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.04716796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5130, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.1995472013950348, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.021454328577965497, + "grad_norm": 0.13386963307857513, + "kept_token_count": 655.0234375, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9898921430110932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010107845440506935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.7084197998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.12070922851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5140, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19894541800022125, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.019944737665355206, + "grad_norm": 0.10924848914146423, + "kept_token_count": 655.2296875, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9904278635978698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009572135843336582, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.27215576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0043548583984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5150, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.20002276599407195, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.019074851367622613, + "grad_norm": 0.1032228171825409, + "kept_token_count": 654.815625, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.989986264705658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010013729613274337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.9599182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4650390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5160, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.19994817972183226, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.017201073560863735, + "grad_norm": 0.12593962252140045, + "kept_token_count": 654.665625, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9908962368965148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009103762172162532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.164990234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6449035644531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5170, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.19962425380945206, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.01541329137980938, + "grad_norm": 0.08073895424604416, + "kept_token_count": 654.9265625, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9903998672962189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009600123949348926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.34239807128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.29759521484374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5180, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.1999776318669319, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.019811070803552867, + "grad_norm": 0.10952325910329819, + "kept_token_count": 654.6703125, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9904564559459687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009543556161224842, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.66835021972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.75840759277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5190, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.19950381219387053, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.01667270911857486, + "grad_norm": 0.129144087433815, + "kept_token_count": 655.0703125, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9905923545360565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009407645370811225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.3349243164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.03192443847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5200, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1987643525004387, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.017337990831583737, + "grad_norm": 0.12757056951522827, + "kept_token_count": 655.4875, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9908486723899841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009151336830109357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.0866302490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6607147216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5210, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.199648018181324, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.01739072548225522, + "grad_norm": 0.133268341422081, + "kept_token_count": 654.925, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9907777607440948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009222244191914798, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.32745666503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.081005859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5220, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2002253606915474, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.018393751699477434, + "grad_norm": 0.14204122126102448, + "kept_token_count": 654.3671875, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9904761493206025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009523848723620176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.5188323974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1591766357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5230, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20071610510349275, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.016860597766935825, + "grad_norm": 0.16849562525749207, + "kept_token_count": 654.025, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9902076542377471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00979235302656889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.4419738769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.38929443359376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5240, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.1991618424654007, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.01785425320267677, + "grad_norm": 0.1151302307844162, + "kept_token_count": 655.603125, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9909827530384063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009017254412174224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.5956207275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5408966064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5250, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.1987851783633232, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.020079537015408277, + "grad_norm": 0.16784228384494781, + "kept_token_count": 655.66875, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9903581202030182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009641868155449628, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.66624450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9931335449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5260, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2004711374640465, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.017534630186855794, + "grad_norm": 0.13865163922309875, + "kept_token_count": 654.434375, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910084009170532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0089915887452662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.9367279052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7791809082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5270, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.19992428570985793, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.017806959990411996, + "grad_norm": 0.11602089554071426, + "kept_token_count": 654.4390625, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9906969726085663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009303020220249892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.44899291992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.40639343261716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5280, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20094911456108094, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.016513519268482924, + "grad_norm": 0.10844706743955612, + "kept_token_count": 653.7828125, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9904583692550659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009541645180433989, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.34368896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.75547790527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5290, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20018875151872634, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.017024540062993764, + "grad_norm": 0.1172213926911354, + "kept_token_count": 654.6765625, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.990431147813797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009568861220031977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.7273284912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.12035217285154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5300, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19891058057546615, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.016560484748333694, + "grad_norm": 0.08176779001951218, + "kept_token_count": 655.3875, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9909590363502503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040950890630483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.3245788574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8223602294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5310, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19879017621278763, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.01841042311862111, + "grad_norm": 0.10663530230522156, + "kept_token_count": 655.8375, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9903012573719024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009698743745684624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.8046630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1180725097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5320, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.1994119256734848, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.016537673864513637, + "grad_norm": 0.1187060996890068, + "kept_token_count": 655.2515625, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.990670919418335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009329071454703807, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.08910217285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.19182739257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19992803782224655, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.01655857227742672, + "grad_norm": 0.13757239282131195, + "kept_token_count": 654.6625, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9905308902263641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009469120856374502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.03192138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7678680419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5340, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19997647553682327, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.016149712074548007, + "grad_norm": 0.09765104204416275, + "kept_token_count": 654.7734375, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9910124659538269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008987549226731062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.1939361572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.28887939453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5350, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20044454634189607, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.017135225515812636, + "grad_norm": 0.1472192257642746, + "kept_token_count": 654.1515625, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9899282574653625, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010071747191250324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.7234771728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5360, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20050088763237, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.016120705287903547, + "grad_norm": 0.11526899039745331, + "kept_token_count": 654.1078125, + "learning_rate": 2.865638159836098e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9902580916881562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0097418955527246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.4952423095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5696594238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5370, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.20005493760108947, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.01811309177428484, + "grad_norm": 0.12643414735794067, + "kept_token_count": 654.478125, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906343996524811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009365591779351234, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.34682312011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.76820373535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20022973865270616, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.018864054512232543, + "grad_norm": 0.1184534952044487, + "kept_token_count": 654.3046875, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9901779055595398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00982207916676998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.42908935546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.04253540039065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2007067769765854, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.01822269558906555, + "grad_norm": 0.12987078726291656, + "kept_token_count": 654.3390625, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.990696132183075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009303861856460571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.2877899169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5725128173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5400, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.19930016547441481, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.017187405843287706, + "grad_norm": 0.11048973351716995, + "kept_token_count": 655.175, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9905237197875977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009476274624466896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.0552551269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.220654296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19934428632259368, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.018113253358751535, + "grad_norm": 0.11657912284135818, + "kept_token_count": 655.0359375, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906406760215759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009359332639724017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.072900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.88469848632815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5420, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.2000044599175453, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.016487739235162734, + "grad_norm": 0.14066995680332184, + "kept_token_count": 654.571875, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9906440317630768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009355955850332975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.4681396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9917785644531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5430, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.20003971606492996, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.015921991597861052, + "grad_norm": 0.12574627995491028, + "kept_token_count": 654.5703125, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9911231815814971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00887681245803833, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.6170623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0349548339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5440, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20076428353786469, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.016037168074399234, + "grad_norm": 0.15778091549873352, + "kept_token_count": 653.9515625, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9903938829898834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00960611617192626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.63475646972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.65273132324216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5450, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20000765323638917, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.015612719859927893, + "grad_norm": 0.13441504538059235, + "kept_token_count": 654.703125, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9907809495925903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009219045657664538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.5087463378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.69642944335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5460, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20009645372629165, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.01745077082887292, + "grad_norm": 0.09555301815271378, + "kept_token_count": 654.4453125, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9902023851871491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009797614999115467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.87464904785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8024139404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5470, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.2004389986395836, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.017316310200840235, + "grad_norm": 0.11509184539318085, + "kept_token_count": 654.2765625, + "learning_rate": 2.858013995286594e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9905731797218322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009426816925406456, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.3612518310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.01227416992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19913101345300674, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.017774035222828388, + "grad_norm": 0.14200422167778015, + "kept_token_count": 655.434375, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9905238389968872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009476164355874062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.30906066894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1179718017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5490, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20086744129657746, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.020426125079393388, + "grad_norm": 0.09369434416294098, + "kept_token_count": 654.05625, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9901243507862091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009875637106597423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.4767669677734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0748809814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20074338614940643, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.015969719830900432, + "grad_norm": 0.12430235743522644, + "kept_token_count": 653.8921875, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9905235588550567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009476453345268965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.5628662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8217529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5510, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.2007138729095459, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.017277094535529613, + "grad_norm": 0.09930719435214996, + "kept_token_count": 653.9375, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909007787704468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009099225793033839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.7757507324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2454437255859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5520, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.19927787333726882, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.015385432075709105, + "grad_norm": 0.1476498395204544, + "kept_token_count": 655.1453125, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9903783917427063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009621629305183888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.5545593261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.22235717773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5530, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19945921152830123, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.01919936891645193, + "grad_norm": 0.15809744596481323, + "kept_token_count": 655.25, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9901469588279724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009853049647063017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.8366302490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.93748474121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5540, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.20016675740480422, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.017856357898563147, + "grad_norm": 0.14126995205879211, + "kept_token_count": 654.325, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9906525552272797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009347442165017128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.58494873046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.83447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5550, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20040056258440017, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.020923668053001165, + "grad_norm": 0.1326000690460205, + "kept_token_count": 654.59375, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9903501510620117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009649846330285072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.448193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1822967529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5560, + "valid_token_count": 818.65 + }, + { + "actual_token_mask_ratio": 0.20059030503034592, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01779460860416293, + "grad_norm": 0.12646916508674622, + "kept_token_count": 654.415625, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9910079360008239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008992065116763115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.4050598144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2884033203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5570, + "valid_token_count": 818.6265625 + }, + { + "actual_token_mask_ratio": 0.19940305948257447, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.020940673444420098, + "grad_norm": 0.11413361877202988, + "kept_token_count": 655.4, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9902944207191468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009705593530088662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.4857666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.21535339355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5580, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.19942951798439026, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.017779379058629273, + "grad_norm": 0.11244891583919525, + "kept_token_count": 655.0046875, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9905158340930938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009484171867370605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.4924072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.31430969238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20019859224557876, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.02065817276015878, + "grad_norm": 0.12004351615905762, + "kept_token_count": 654.6578125, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9907067477703094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00929323984310031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.12062377929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.81077880859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5600, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.20049847960472106, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.019361775182187558, + "grad_norm": 0.10627549141645432, + "kept_token_count": 654.3625, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9902045667171478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009795429557561875, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.8212432861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8350372314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2001471221446991, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.01727109756320715, + "grad_norm": 0.10124414414167404, + "kept_token_count": 654.3390625, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909458041191102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009054185915738344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.99913024902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9255828857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5620, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.20068497508764266, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.02053246907889843, + "grad_norm": 0.11493398249149323, + "kept_token_count": 654.09375, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9908895850181579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009110409673303366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.97544555664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8376098632813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5630, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.19886801540851592, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.016054777521640064, + "grad_norm": 0.14187504351139069, + "kept_token_count": 655.50625, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9904650866985321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009534898586571217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.34208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8159912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5640, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2003805622458458, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.019598728232085704, + "grad_norm": 0.127269446849823, + "kept_token_count": 654.53125, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.98969207406044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010307905450463294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.3893707275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.44919738769534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5650, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.20033404380083084, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.01701321844011545, + "grad_norm": 0.12826594710350037, + "kept_token_count": 654.178125, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9896560251712799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010343976505100726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.7487091064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0486541748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5660, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20026192516088487, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.01979239033535123, + "grad_norm": 0.14953060448169708, + "kept_token_count": 654.234375, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9904098391532898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009590161871165038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.4204986572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.13030395507815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5670, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.20045698583126068, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.015776912961155177, + "grad_norm": 0.1410495638847351, + "kept_token_count": 654.2296875, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9904161989688873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009583807457238436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.7610290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.317919921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5680, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.1992128387093544, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.01974242003634572, + "grad_norm": 0.14201179146766663, + "kept_token_count": 655.1375, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.990018618106842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009981398563832044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.03545837402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9991851806641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5690, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2001170128583908, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.014823016244918108, + "grad_norm": 0.09004012495279312, + "kept_token_count": 654.596875, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9902719795703888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00972802536562085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.02088928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9429229736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5700, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20091042220592498, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.01984606795012951, + "grad_norm": 0.11087047308683395, + "kept_token_count": 654.115625, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9904533207416535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009546675067394973, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.7462860107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7061462402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5710, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.19970069825649261, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.018770266510546207, + "grad_norm": 0.1140713319182396, + "kept_token_count": 654.9015625, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9908593535423279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009140645340085029, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.7522735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.17225036621096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.20086439996957778, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.022117822617292403, + "grad_norm": 0.1318981945514679, + "kept_token_count": 653.7421875, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0269, + "mgd_cos_sim": 0.9903680860996247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009631911665201187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.16045837402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2445770263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5730, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20101465433835983, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.018540462013334036, + "grad_norm": 0.14119407534599304, + "kept_token_count": 653.6578125, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9906628251075744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009337171446532011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.85474853515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7863342285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20062750428915024, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.016678918991237877, + "grad_norm": 0.11432802677154541, + "kept_token_count": 654.23125, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9907885789871216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009211427997797728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.01898193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3415863037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5750, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20050129443407058, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.015761073771864177, + "grad_norm": 0.12955529987812042, + "kept_token_count": 653.984375, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9907789409160614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009221053775399923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.9614685058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8827819824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5760, + "valid_token_count": 817.9921875 + }, + { + "actual_token_mask_ratio": 0.19938897490501403, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.017413049284368754, + "grad_norm": 0.09849334508180618, + "kept_token_count": 654.9875, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912687301635742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008731278963387013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.5082305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.0913940429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5770, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19931131601333618, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.019214657414704562, + "grad_norm": 0.12129244953393936, + "kept_token_count": 655.103125, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9906687140464783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009331290423870087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.77940673828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5167724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5780, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.20048314034938813, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.015603936836123466, + "grad_norm": 0.1000269204378128, + "kept_token_count": 654.44375, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.991032338142395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00896766409277916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3651092529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3966888427734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5790, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.20039046853780745, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.016222918312996625, + "grad_norm": 0.11595045775175095, + "kept_token_count": 654.1671875, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9907026410102844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009297362249344587, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.8848907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3720703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5800, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.19967909902334213, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.01644203197211027, + "grad_norm": 0.13494311273097992, + "kept_token_count": 654.7609375, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9905198693275452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009480120614171028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.3512329101562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.94605407714846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5810, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2001541867852211, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.015823709778487682, + "grad_norm": 0.11103705316781998, + "kept_token_count": 654.40625, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9906855702400208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009314426779747009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.53464660644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7752197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5820, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.20125871151685715, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.01712657595053315, + "grad_norm": 0.1021684780716896, + "kept_token_count": 653.4421875, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9909509897232056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009049010276794434, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.51267395019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.03086853027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5830, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19946926236152648, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.01648227348923683, + "grad_norm": 0.1316683441400528, + "kept_token_count": 655.0578125, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9908206641674042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009179327916353942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.1111145019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.22706604003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5840, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20071645230054855, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.017123069800436496, + "grad_norm": 0.11330872029066086, + "kept_token_count": 653.9046875, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907222509384155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009277739562094212, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.4839782714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.85105285644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5850, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.19976287186145783, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.016531973145902157, + "grad_norm": 0.10266707837581635, + "kept_token_count": 654.553125, + "learning_rate": 2.830147042037172e-05, + "loss": 0.021, + "mgd_cos_sim": 0.99110067486763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008899321779608727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.50304260253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.43528137207034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5860, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.2001626282930374, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.016449097264558076, + "grad_norm": 0.12067357450723648, + "kept_token_count": 654.3796875, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.990618371963501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938162449747324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.60915222167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.30860290527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5870, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19954944849014283, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.01783640244975686, + "grad_norm": 0.10350602120161057, + "kept_token_count": 654.8609375, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.990462988615036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00953699927777052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.7092529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.89658508300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5880, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.20056019574403763, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.015208930149674416, + "grad_norm": 0.10989784449338913, + "kept_token_count": 654.3421875, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9905231595039368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00947683621197939, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.61566467285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.10536499023436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5890, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.1994134172797203, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.01642381167039275, + "grad_norm": 0.09654425084590912, + "kept_token_count": 655.11875, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9908873200416565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009112687036395072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.29652709960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.62168273925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5900, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.19914343059062958, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.016656912211328746, + "grad_norm": 0.1193893775343895, + "kept_token_count": 655.1, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9910739004611969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008926094602793455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6763031005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2786804199219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.19903017580509186, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.017347859777510168, + "grad_norm": 0.14032675325870514, + "kept_token_count": 655.484375, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9903646349906922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009635372925549746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.8253601074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7630157470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5920, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.19976384341716766, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.019618835393339397, + "grad_norm": 0.17659859359264374, + "kept_token_count": 654.925, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9899437725543976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010056214407086372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.7680267333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.10984497070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5930, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20082495957612992, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.018494844902306795, + "grad_norm": 0.12787321209907532, + "kept_token_count": 653.921875, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9902839899063111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009716019779443742, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.88206176757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.5309143066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5940, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2005945026874542, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.019637399725615977, + "grad_norm": 0.14570610225200653, + "kept_token_count": 654.2203125, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9902752339839935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009724763687700033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.2260284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3464324951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5950, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20014727264642715, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.01687180157750845, + "grad_norm": 0.14688575267791748, + "kept_token_count": 654.73125, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9908657848834992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009134219586849212, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.48948974609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.57506408691404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5960, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.19950082898139954, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.016740274149924517, + "grad_norm": 0.14069819450378418, + "kept_token_count": 655.00625, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9910385012626648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00896150665357709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.45038146972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7849182128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5970, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.1996404528617859, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.015728992223739625, + "grad_norm": 0.13008365035057068, + "kept_token_count": 655.125, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9912115156650543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008788480050861835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.1085205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1826995849609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5980, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.20033873468637467, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.017357072606682778, + "grad_norm": 0.12704585492610931, + "kept_token_count": 654.49375, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9914954423904419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008504555281251668, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.0896881103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6951599121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 5990, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.19971794486045838, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.01635226095095277, + "grad_norm": 0.1456683725118637, + "kept_token_count": 655.1203125, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9905884385108947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009411558136343956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.12896118164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8343200683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6000, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.19979831874370574, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.01851545451208949, + "grad_norm": 0.19626332819461823, + "kept_token_count": 654.5453125, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9908711850643158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009128817170858384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.32139892578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9868408203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6010, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.19992378503084182, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.017294422164559363, + "grad_norm": 0.13132278621196747, + "kept_token_count": 654.678125, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9908307909965515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009169213380664587, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.8380096435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.60614624023435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6020, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.19947535842657088, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.019468637369573116, + "grad_norm": 0.14449195563793182, + "kept_token_count": 654.9390625, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9906683027744293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009331698156893253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.8991455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9927093505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6030, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19985170811414718, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.015587362088263034, + "grad_norm": 0.15246249735355377, + "kept_token_count": 654.4203125, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.990514487028122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0094855141825974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.6623870849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9696380615234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6040, + "valid_token_count": 817.86875 + }, + { + "actual_token_mask_ratio": 0.20055498778820038, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01783415200188756, + "grad_norm": 0.11645980924367905, + "kept_token_count": 654.265625, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9906347870826722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009365215618163347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.16188659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7242736816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6050, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19910406023263932, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.017013540863990782, + "grad_norm": 0.12713022530078888, + "kept_token_count": 655.4625, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9901289522647858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009871058631688356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.57508544921876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7602905273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6060, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19934562146663665, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.016803179867565632, + "grad_norm": 0.09385703504085541, + "kept_token_count": 655.384375, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9905944228172302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009405552688986064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.5662811279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8654418945313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6070, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.1995173215866089, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.021015966311097145, + "grad_norm": 0.12138386815786362, + "kept_token_count": 654.7546875, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9910444140434265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008955596387386322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.8853729248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8375244140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6080, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.20014275312423707, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.017959040869027377, + "grad_norm": 0.14700280129909515, + "kept_token_count": 654.35625, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9905628204345703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009437175933271646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.57256469726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0167602539062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6090, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19953683614730836, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.01745800403878093, + "grad_norm": 0.12814709544181824, + "kept_token_count": 655.153125, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9911844074726105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008815592247992754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.9253356933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.24485778808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6100, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19958158284425737, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.01960740415379405, + "grad_norm": 0.13966256380081177, + "kept_token_count": 655.48125, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9903782546520233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009621748980134726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8136779785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1026885986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6110, + "valid_token_count": 818.9234375 + }, + { + "actual_token_mask_ratio": 0.1994232565164566, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.01834684582427144, + "grad_norm": 0.13388770818710327, + "kept_token_count": 655.1421875, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9906767189502717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009323271177709102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.83896789550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4281494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6120, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19951315522193908, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.018560083210468294, + "grad_norm": 0.104826919734478, + "kept_token_count": 654.809375, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9908157765865326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009184216428548098, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.43125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5941864013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6130, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.1994483083486557, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.018549940455704926, + "grad_norm": 0.15109696984291077, + "kept_token_count": 654.9140625, + "learning_rate": 2.808117394745975e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9911415457725525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008858454879373312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.0417022705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.978955078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6140, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20050784796476365, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.015722368843853474, + "grad_norm": 0.10988083481788635, + "kept_token_count": 654.2265625, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9909915864467621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009008409082889557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.33494873046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6903991699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6150, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20064717382192612, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.01766856899484992, + "grad_norm": 0.1473887860774994, + "kept_token_count": 653.9375, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.989894300699234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0101057056337595, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.8480560302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3892456054688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.1996584340929985, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.015220771916210652, + "grad_norm": 0.11412596702575684, + "kept_token_count": 654.8828125, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9909918427467346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900815473869443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.7184539794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1609741210938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6170, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2001751959323883, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.015994585026055573, + "grad_norm": 0.14951783418655396, + "kept_token_count": 654.590625, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9902720749378204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009727930650115013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.65001525878904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0907501220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6180, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20160536766052245, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.01707663768902421, + "grad_norm": 0.12700524926185608, + "kept_token_count": 653.2015625, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.991067373752594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008932612277567387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.4265502929687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2433624267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6190, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20142488330602645, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.01731910677626729, + "grad_norm": 0.14855796098709106, + "kept_token_count": 653.5203125, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9911196887493133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008880302123725415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.2751983642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7674041748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6200, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.2006383016705513, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.01712823109701276, + "grad_norm": 0.09919918328523636, + "kept_token_count": 653.9828125, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.991142737865448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008857260458171368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.62538757324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9526702880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6210, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.1992362394928932, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.015905092097818853, + "grad_norm": 0.09441491961479187, + "kept_token_count": 655.221875, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9908988058567048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009101189021021128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.3630828857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4119079589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6220, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20090881288051604, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.016524951066821815, + "grad_norm": 0.14503370225429535, + "kept_token_count": 653.9859375, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9902247726917267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009775227680802346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.1943420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.27919616699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20008436292409898, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.017595008946955204, + "grad_norm": 0.1290299892425537, + "kept_token_count": 654.3046875, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9913932144641876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008606782648712397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.84336853027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.49818115234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6240, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.19976603090763093, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.016380946431308985, + "grad_norm": 0.13269703090190887, + "kept_token_count": 654.9125, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9913008630275726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008699139021337033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.62513732910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.98956604003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6250, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20006811022758483, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.01660795547068119, + "grad_norm": 0.13167035579681396, + "kept_token_count": 654.64375, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912128329277039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008787171263247729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0801300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.42547912597655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6260, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19966306537389755, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.015884715877473355, + "grad_norm": 0.1381085366010666, + "kept_token_count": 654.7765625, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9913566470146179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008643343951553107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.46612548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3495727539063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6270, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.20063869953155516, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.0152560212649405, + "grad_norm": 0.12084293365478516, + "kept_token_count": 653.8921875, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9907529950141907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009246999397873878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.7505279541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1399627685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6280, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2007686823606491, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.018251576740294696, + "grad_norm": 0.1236145868897438, + "kept_token_count": 654.25, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9903564989566803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009643492009490728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.172998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2950439453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6290, + "valid_token_count": 818.596875 + }, + { + "actual_token_mask_ratio": 0.199311001598835, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.016913298424333333, + "grad_norm": 0.10302140563726425, + "kept_token_count": 655.1234375, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9911415934562683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008858408778905869, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.02447814941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6591766357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6300, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.2000850111246109, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.01574571691453457, + "grad_norm": 0.1190619096159935, + "kept_token_count": 654.6296875, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9912671983242035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873280456289649, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.9458740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8140472412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6310, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19957370311021805, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.017207894660532474, + "grad_norm": 0.11401627957820892, + "kept_token_count": 654.9765625, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9906999886035919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009300014562904834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.1764404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.37485656738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6320, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19944602996110916, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.018510486278682948, + "grad_norm": 0.13271884620189667, + "kept_token_count": 654.8859375, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.99039426445961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009605717472732067, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.99876708984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1003814697266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6330, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.20030927062034606, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.016433572210371495, + "grad_norm": 0.13509802520275116, + "kept_token_count": 654.5765625, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9905423164367676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00945768840610981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.88429565429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3463989257813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6340, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.19968314915895463, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.020108845084905624, + "grad_norm": 0.11705353856086731, + "kept_token_count": 654.825, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.990954452753067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009045547153800727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.71739807128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.988671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.19909163564443588, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.016013731807470323, + "grad_norm": 0.12055867165327072, + "kept_token_count": 655.36875, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9904738247394562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009526169300079346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.3060028076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9187438964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6360, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20059116780757905, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.016661742981523276, + "grad_norm": 0.13275302946567535, + "kept_token_count": 654.084375, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9910936713218689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008906323276460171, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.1770080566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.17784118652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6370, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.20020715296268463, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.01688061896711588, + "grad_norm": 0.11452330648899078, + "kept_token_count": 654.834375, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9901475250720978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009852472133934498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.4855590820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.74773559570315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6380, + "valid_token_count": 818.7484375 + }, + { + "actual_token_mask_ratio": 0.2007594734430313, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.01822515968233347, + "grad_norm": 0.11684684455394745, + "kept_token_count": 653.965625, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.990709912776947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009290091134607792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.5047637939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2063842773438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6390, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2002631261944771, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.019685081858187913, + "grad_norm": 0.14865699410438538, + "kept_token_count": 654.340625, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9899416744709015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010058322176337241, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.07745666503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6529113769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6400, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20043940991163253, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.019022608548402785, + "grad_norm": 0.1676410436630249, + "kept_token_count": 654.25, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9911436438560486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00885636117309332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.4284729003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.60141906738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6410, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19988000243902207, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.018325382377952337, + "grad_norm": 0.1121663898229599, + "kept_token_count": 654.578125, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908184766769409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009181535057723521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.7576141357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7491485595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6420, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.1998242110013962, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.01673069829121232, + "grad_norm": 0.12713614106178284, + "kept_token_count": 654.6953125, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9905797064304351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009420288447290658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.1745147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2953277587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6430, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.1993319571018219, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.018686455860733985, + "grad_norm": 0.12317612767219543, + "kept_token_count": 655.0953125, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9901947975158691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009805199503898621, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.6062927246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3829071044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6440, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19998109340667725, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.01807704595848918, + "grad_norm": 0.11265867203474045, + "kept_token_count": 654.5453125, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9907934188842773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009206582978367805, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.7459381103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0393493652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6450, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.19992488473653794, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.018155492283403872, + "grad_norm": 0.16257746517658234, + "kept_token_count": 654.634375, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9909372806549073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009062719997018576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.0425323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5063751220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6460, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2001340851187706, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.015826415456831455, + "grad_norm": 0.12243740260601044, + "kept_token_count": 654.63125, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9910288453102112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008971155900508166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.2774261474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.36228332519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6470, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.20130923986434937, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.016719064209610224, + "grad_norm": 0.13287237286567688, + "kept_token_count": 653.703125, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9912522375583649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008747771196067334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.31346435546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7222412109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6480, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.199823297560215, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.019394761975854637, + "grad_norm": 0.10994557291269302, + "kept_token_count": 654.7875, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9909199059009552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009080082923173905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.99464416503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.08424377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6490, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20020537227392196, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.018220920767635106, + "grad_norm": 0.12962529063224792, + "kept_token_count": 654.275, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9904161930084229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009583801776170731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.54072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.61407165527345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6500, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.1992333322763443, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.019201909936964513, + "grad_norm": 0.11108897626399994, + "kept_token_count": 655.2390625, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.990609097480774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009390916116535664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.53667907714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4450164794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6510, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.1994275376200676, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.017407154012471438, + "grad_norm": 0.09485536068677902, + "kept_token_count": 655.1109375, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9905633211135865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009436670411378145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.8583099365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.45693054199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6520, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20080383270978927, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.017780703771859408, + "grad_norm": 0.12212402373552322, + "kept_token_count": 653.6109375, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9901121139526368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009887896664440631, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.99150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.63376770019534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6530, + "valid_token_count": 817.8359375 + }, + { + "actual_token_mask_ratio": 0.20020245164632797, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.015294706728309394, + "grad_norm": 0.12024403363466263, + "kept_token_count": 654.440625, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9907443821430206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009255611430853605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.93811950683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1161376953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6540, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19943997859954835, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.016862767562270165, + "grad_norm": 0.11123422533273697, + "kept_token_count": 654.7578125, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9911173939704895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008882623910903931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.38840637207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.26805419921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6550, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.19966302514076234, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.017058449331671, + "grad_norm": 0.1204473152756691, + "kept_token_count": 654.996875, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9905562043190003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009443806950002909, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.0402374267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.41176147460936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6560, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19999646842479707, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.017731614504009487, + "grad_norm": 0.10510517656803131, + "kept_token_count": 654.6671875, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9905842661857605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009415739215910435, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.40123596191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2270935058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6570, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19991289377212523, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.015508454758673906, + "grad_norm": 0.10457105934619904, + "kept_token_count": 654.690625, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9898814141750336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010118593741208316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.5724304199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.55751647949216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6580, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19981119483709336, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.01783238835632801, + "grad_norm": 0.1495552957057953, + "kept_token_count": 654.496875, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9912348389625549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008765157405287027, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.6392120361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6000244140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6590, + "valid_token_count": 817.9234375 + }, + { + "actual_token_mask_ratio": 0.20109607577323912, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.018842928949743508, + "grad_norm": 0.13035115599632263, + "kept_token_count": 653.809375, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9900534749031067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009946530312299728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.07249450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.74794006347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6600, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20146357864141465, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.018441309221088886, + "grad_norm": 0.10627862066030502, + "kept_token_count": 653.3109375, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9906103491783143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009389658458530903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.8234344482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.26398315429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.19999614357948303, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.019704925920814277, + "grad_norm": 0.15594927966594696, + "kept_token_count": 654.571875, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9902760744094848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009723913948982954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.03215942382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.13031005859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6620, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.19927199929952621, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.01978625189512968, + "grad_norm": 0.1269637495279312, + "kept_token_count": 655.1703125, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.991119521856308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00888048391789198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.65308837890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0445129394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6630, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.2001440480351448, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.019415105506777764, + "grad_norm": 0.13078539073467255, + "kept_token_count": 654.790625, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9904526710510254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009547324385493994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.74323120117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.33929748535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6640, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20038006007671355, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.021518640872091054, + "grad_norm": 0.12104244530200958, + "kept_token_count": 654.315625, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9904886364936829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009511360712349416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5727905273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.78050537109374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6650, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.19982904344797134, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.016087057068943977, + "grad_norm": 0.16711488366127014, + "kept_token_count": 654.7, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9904523253440857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009547673352062701, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.26231689453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7425140380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6660, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19973969906568528, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.018962070532143115, + "grad_norm": 0.1195007860660553, + "kept_token_count": 654.796875, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9903304159641266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009669576957821845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.3272277832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7420288085938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6670, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.19955285191535949, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.014493829663842916, + "grad_norm": 0.096594899892807, + "kept_token_count": 654.8890625, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9908826112747192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009117377549409866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.7899963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.23634338378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6680, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.198970428109169, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.016033187322318555, + "grad_norm": 0.11448469758033752, + "kept_token_count": 655.3265625, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9912213385105133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008778646681457758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.0882293701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.03017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6690, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.19966956973075867, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.017166668828576804, + "grad_norm": 0.11512381583452225, + "kept_token_count": 654.996875, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907898843288422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009210121538490056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.5654266357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9618621826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6700, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19989169538021087, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.019464630633592606, + "grad_norm": 0.1214587613940239, + "kept_token_count": 654.859375, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9898707032203674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010129313636571168, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.8722351074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.99280700683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6710, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.20070981681346894, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.02090449659153819, + "grad_norm": 0.12426464259624481, + "kept_token_count": 654.3265625, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.990779447555542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009220545087009669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.9852569580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5812194824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.1999659314751625, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.018291372805833817, + "grad_norm": 0.13048522174358368, + "kept_token_count": 654.6640625, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908547103404999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009145296458154917, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.34410400390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1320404052734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6730, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.19970404505729675, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.0166908361017704, + "grad_norm": 0.11158086359500885, + "kept_token_count": 654.6109375, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9904367446899414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009563256893306971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.5226654052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4150024414063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6740, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19961017817258836, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.02005609469488263, + "grad_norm": 0.11499522626399994, + "kept_token_count": 655.103125, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9906201481819152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00937984436750412, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.8641815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.35487670898436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6750, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.1996747925877571, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.018812856636941433, + "grad_norm": 0.13941305875778198, + "kept_token_count": 654.85625, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9909926116466522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009007382113486529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.2364532470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.08672485351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6760, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2001571238040924, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.01825677854940295, + "grad_norm": 0.1312652975320816, + "kept_token_count": 654.48125, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9904895126819611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009510489832609891, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.18560791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.03363647460935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6770, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.200839664041996, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.018625089153647422, + "grad_norm": 0.124686598777771, + "kept_token_count": 653.74375, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9907325267791748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00926747526973486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.65472412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1340301513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6780, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.19977944493293762, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.01790870577096939, + "grad_norm": 0.12405917793512344, + "kept_token_count": 654.80625, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9906907856464386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009309225343167781, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.97991638183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7171264648438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6790, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.20000331699848176, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.017628161888569593, + "grad_norm": 0.13431301712989807, + "kept_token_count": 654.528125, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9906002163887024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0093997772783041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.74364318847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6671142578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6800, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20011284798383713, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.019014917593449355, + "grad_norm": 0.11689522117376328, + "kept_token_count": 654.7921875, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9905535936355591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009446410089731216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.820556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8412384033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6810, + "valid_token_count": 818.6046875 + }, + { + "actual_token_mask_ratio": 0.20039789974689484, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.019678907468914986, + "grad_norm": 0.11310361325740814, + "kept_token_count": 654.1546875, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9899583578109741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010041657090187072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.75901794433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2231048583984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6820, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20035200864076613, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.01617771266028285, + "grad_norm": 0.10857000946998596, + "kept_token_count": 654.1296875, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9911031246185302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008896857127547264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.25050354003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2609100341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6830, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.20028171092271804, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.01825512582436204, + "grad_norm": 0.12038035690784454, + "kept_token_count": 654.4734375, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.990122240781784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009877755865454673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.4747802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0695098876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6840, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.20041076987981796, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.016869428660720588, + "grad_norm": 0.1140720546245575, + "kept_token_count": 654.1734375, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9906440138816833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009355981554836034, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.3147766113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0476379394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6850, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.19918373823165894, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.016730744391679764, + "grad_norm": 0.11778765916824341, + "kept_token_count": 655.3421875, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9905659198760987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009434068296104669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8824798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4959991455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6860, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19972380846738816, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.01703343354165554, + "grad_norm": 0.14120809733867645, + "kept_token_count": 654.815625, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909641683101654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009035830665379763, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.30385131835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8146484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6870, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19936641305685043, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.017888775654137136, + "grad_norm": 0.14830073714256287, + "kept_token_count": 655.1375, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913239479064941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008676056191325188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.73931579589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9261444091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6880, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.19961071014404297, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.016867329832166433, + "grad_norm": 0.10040957480669022, + "kept_token_count": 654.778125, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.991023188829422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008976818714290857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.9623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.81556396484376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6890, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.19981518834829332, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.017678300943225624, + "grad_norm": 0.11980205029249191, + "kept_token_count": 654.878125, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9913815200328827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008618466928601265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.8170562744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.32334289550784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6900, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20015523582696915, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.018972117360681294, + "grad_norm": 0.13382963836193085, + "kept_token_count": 654.3703125, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9899835050106048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010016510635614395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.8873718261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6836639404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6910, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20002047270536422, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01731772255152464, + "grad_norm": 0.10970032215118408, + "kept_token_count": 654.475, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9911776959896088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008822315279394389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.9265167236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3902099609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6920, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.1992414712905884, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.018444565031677483, + "grad_norm": 0.1339658796787262, + "kept_token_count": 655.3640625, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9909648835659027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009035126026719808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.65443420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6701202392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6930, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.19946937412023544, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.016379085555672645, + "grad_norm": 0.10762582719326019, + "kept_token_count": 655.20625, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9903742015361786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0096257827244699, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.2877532958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0325042724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6940, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.19927351474761962, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.018616127781569958, + "grad_norm": 0.13043837249279022, + "kept_token_count": 655.3296875, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9913195610046387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008680445700883865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9895477294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.49891357421876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6950, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19975797832012177, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.017986715026199817, + "grad_norm": 0.13611651957035065, + "kept_token_count": 654.78125, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9910744369029999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00892558041960001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.98158264160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3418670654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6960, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.2000991627573967, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.01798937050625682, + "grad_norm": 0.09464650601148605, + "kept_token_count": 654.584375, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9907522559165954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009247750882059336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.4912353515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9110931396484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20056440085172653, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.018472669832408427, + "grad_norm": 0.10451134294271469, + "kept_token_count": 654.3671875, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9912898182868958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087101798504591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.9138427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1869720458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6980, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.19970347732305527, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.01722909277305007, + "grad_norm": 0.12529855966567993, + "kept_token_count": 654.8078125, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907756328582764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009224371798336507, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.41001892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5528228759766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 6990, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19992536008358003, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.01644656592980027, + "grad_norm": 0.11011704057455063, + "kept_token_count": 654.85, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9903027474880218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009697255492210389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.8807067871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1963623046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7000, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20028412491083145, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01889811912551522, + "grad_norm": 0.1186550110578537, + "kept_token_count": 654.2578125, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9906123340129852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009387669712305069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.61748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.70098876953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7010, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20009187012910842, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.01614140160381794, + "grad_norm": 0.08623448759317398, + "kept_token_count": 654.6125, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9910502672195435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008949723932892084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.7114776611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.22214660644534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7020, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19936344772577286, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.01595936333760619, + "grad_norm": 0.13844259083271027, + "kept_token_count": 655.1859375, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9902112245559692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009788763709366322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.2208251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5558135986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7030, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.199779212474823, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.01767143476754427, + "grad_norm": 0.10460204631090164, + "kept_token_count": 654.715625, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.991042536497116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008957464713603258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.46913452148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7530914306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7040, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19928252547979355, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.01741414489224553, + "grad_norm": 0.12299251556396484, + "kept_token_count": 654.984375, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9910016119480133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008998392429202796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.39588317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.13404541015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7050, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.2001274362206459, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.0159392979927361, + "grad_norm": 0.17368943989276886, + "kept_token_count": 654.4046875, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9911915063858032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008808494731783866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.55938110351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0710144042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7060, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.20049366354942322, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.019148987717926502, + "grad_norm": 0.12007144838571548, + "kept_token_count": 654.3203125, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9902289569377899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009771033190190792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.4582061767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.64178771972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7070, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.19993606209754944, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.01577825527638197, + "grad_norm": 0.11806377023458481, + "kept_token_count": 654.44375, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9910857677459717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008914237096905708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.3039215087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.72862243652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7080, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20022235810756683, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.01730042416602373, + "grad_norm": 0.1576802283525467, + "kept_token_count": 654.5765625, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9913244724273682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008675517421215773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.3224334716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.79469299316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7090, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.19982475340366362, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.016654570028185843, + "grad_norm": 0.12929031252861023, + "kept_token_count": 654.675, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9903743684291839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00962561583146453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.1145538330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.57677307128904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7100, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19989010244607924, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.01789351338520646, + "grad_norm": 0.09001969546079636, + "kept_token_count": 654.5515625, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9912622213363648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873778685927391, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.9154876708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.87144165039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7110, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.1998589038848877, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.017601931001991034, + "grad_norm": 0.13307535648345947, + "kept_token_count": 654.6265625, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9908523321151733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009147650469094515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.1660919189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.38898010253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7120, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19988076835870744, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.017853297106921674, + "grad_norm": 0.10518141090869904, + "kept_token_count": 654.5890625, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9913401186466217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0086598826572299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.73797607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8322052001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7130, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19886561781167983, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.016617361549288036, + "grad_norm": 0.13974225521087646, + "kept_token_count": 655.4546875, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9909995377063752, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009000459313392639, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4528503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.51470642089845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20027923732995986, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.014850085787475108, + "grad_norm": 0.11304163932800293, + "kept_token_count": 654.3296875, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9909250557422637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009074936900287867, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.00599975585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1151092529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7150, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20085730105638505, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.017354735266417266, + "grad_norm": 0.12101753801107407, + "kept_token_count": 653.975, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9910142719745636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0089857273735106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.4958099365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6098663330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7160, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2002449095249176, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.018665943201631306, + "grad_norm": 0.12113799899816513, + "kept_token_count": 654.6265625, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9910917818546295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008908226434141398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.7614501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9613098144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7170, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19972779601812363, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.017473798524588347, + "grad_norm": 0.13190296292304993, + "kept_token_count": 654.890625, + "learning_rate": 2.715563197227934e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910360991954803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008963899686932564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.88030395507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1629333496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7180, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19976565092802048, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.017563598044216634, + "grad_norm": 0.11092185229063034, + "kept_token_count": 655.11875, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9907790124416351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009220988862216473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.11436767578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0463562011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7190, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.1999628409743309, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.016056278627365826, + "grad_norm": 0.10093126446008682, + "kept_token_count": 654.6953125, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9907827556133271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009217238519340754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.9134460449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.01043395996095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7200, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.19963668435811996, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.01720239408314228, + "grad_norm": 0.11009865254163742, + "kept_token_count": 654.8234375, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.990545117855072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009454887732863427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3372467041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.99744567871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7210, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.200863279402256, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015492098499089479, + "grad_norm": 0.11111868917942047, + "kept_token_count": 654.1, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9904979348182679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009502053819596767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3867950439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5408966064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7220, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20008411705493928, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.018246675748378037, + "grad_norm": 0.11336144059896469, + "kept_token_count": 654.415625, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9912872850894928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00871271314099431, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.6052581787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3258941650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7230, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.19909121990203857, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.01932599423453212, + "grad_norm": 0.12377559393644333, + "kept_token_count": 655.4328125, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9911798894405365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00882010767236352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8800872802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9616302490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7240, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20099994391202927, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.015164662804454564, + "grad_norm": 0.08819384127855301, + "kept_token_count": 653.925, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9916854679584504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008314531203359366, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.63243713378904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2489379882812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7250, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20012690722942353, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.01561936903744936, + "grad_norm": 0.10921583324670792, + "kept_token_count": 654.5015625, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9911882817745209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008811699599027634, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8198272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.21437683105466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7260, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.1999972015619278, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.018168373219668865, + "grad_norm": 0.11428358405828476, + "kept_token_count": 654.4734375, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9907154560089111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009284556843340397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.1293975830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.01893005371096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7270, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2000310555100441, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.01795374760404229, + "grad_norm": 0.14339978992938995, + "kept_token_count": 654.6703125, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9906224191188813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009377573151141405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.0285247802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0678649902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7280, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19881186485290528, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.018836649879813195, + "grad_norm": 0.12752704322338104, + "kept_token_count": 655.4875, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.990358418226242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009641585685312749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3628723144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7880523681641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7290, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.1997624382376671, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.016769907716661692, + "grad_norm": 0.11358730494976044, + "kept_token_count": 654.765625, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9915602445602417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008439762331545354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.9985321044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2143859863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7300, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19988456666469573, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.018147848546504974, + "grad_norm": 0.1286490112543106, + "kept_token_count": 654.871875, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9911345958709716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008865415863692761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.04441223144534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.34918518066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7310, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.19933395832777023, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.019522468466311692, + "grad_norm": 0.11885669827461243, + "kept_token_count": 655.0625, + "learning_rate": 2.701855985788514e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9910346508026123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008965353947132826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7802368164063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4201690673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7320, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19995970726013185, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.01654850775375962, + "grad_norm": 0.11775127798318863, + "kept_token_count": 654.34375, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9906539261341095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009346059709787368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.26372985839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1577850341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7330, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.19965792298316956, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.01752565959468484, + "grad_norm": 0.12713809311389923, + "kept_token_count": 655.0578125, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.98993279337883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010067203175276518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.30823974609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.79347839355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7340, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.1998573511838913, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.01640586378052831, + "grad_norm": 0.10832956433296204, + "kept_token_count": 654.85625, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9904100358486175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009589944873005152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.5517547607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4095031738281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.200126776099205, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.021419602446258068, + "grad_norm": 0.13962902128696442, + "kept_token_count": 654.475, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0264, + "mgd_cos_sim": 0.9900120913982391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998791279271245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.99827575683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9236145019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7360, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.1999574512243271, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.020869563054293393, + "grad_norm": 0.13057821989059448, + "kept_token_count": 654.7171875, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9901697218418122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009830267075449229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.4046691894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.07801513671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7370, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.1991698771715164, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.01787819303572178, + "grad_norm": 0.1448535919189453, + "kept_token_count": 655.41875, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9901653051376342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009834698308259249, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.8335845947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.67140197753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7380, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.20077481716871262, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.01941823624074459, + "grad_norm": 0.1511402130126953, + "kept_token_count": 654.1546875, + "learning_rate": 2.694894838415971e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9909038424491883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009096168540418148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.93987426757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.41591796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.19965225756168364, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.01648020576685667, + "grad_norm": 0.12849396467208862, + "kept_token_count": 654.959375, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9905345618724823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009465432725846767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.3359619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1716766357422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7400, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2006103739142418, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.01643838742747903, + "grad_norm": 0.11543378978967667, + "kept_token_count": 653.9921875, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9916403889656067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00835961988195777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0333648681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.92015686035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7410, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19957760721445084, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.01686111595481634, + "grad_norm": 0.1153649389743805, + "kept_token_count": 655.0140625, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9915379703044891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008462032582610846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.2764953613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.16680908203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7420, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19992497563362122, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.01733333682641387, + "grad_norm": 0.10254951566457748, + "kept_token_count": 654.8484375, + "learning_rate": 2.690885071296308e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9906286478042603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00937134837731719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1722839355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.42856750488284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7430, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20003142803907395, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.017243801429867745, + "grad_norm": 0.14086510241031647, + "kept_token_count": 654.5359375, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9911441087722779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008855901286005973, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.98041381835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1508453369141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7440, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.19955661594867707, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.017373621556907894, + "grad_norm": 0.10838709026575089, + "kept_token_count": 655.046875, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.990313982963562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009686018992215394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.2390991210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0049560546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7450, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20029483288526534, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.018545406870543957, + "grad_norm": 0.1297202706336975, + "kept_token_count": 654.1, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912558734416962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008744127862155438, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.46326599121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.02019653320315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7460, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.20034203827381133, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.01800887007266283, + "grad_norm": 0.10243122279644012, + "kept_token_count": 654.159375, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9910213708877563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008978628367185593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.04372253417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.21140747070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7470, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.20092703849077226, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.01603154679760337, + "grad_norm": 0.12258034199476242, + "kept_token_count": 654.090625, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.99091437458992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009085622243583202, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.7751220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6023254394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7480, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.2008368194103241, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015244880877435207, + "grad_norm": 0.0925612598657608, + "kept_token_count": 653.9171875, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9908356666564941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009164347127079964, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.02960815429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.87159423828126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7490, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2001435473561287, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.019036276917904615, + "grad_norm": 0.11903850734233856, + "kept_token_count": 654.5, + "learning_rate": 2.68381231512099e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9901339769363403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009866013005375863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.3425628662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.61570129394534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7500, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.1997060626745224, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.016402472835034133, + "grad_norm": 0.1324203610420227, + "kept_token_count": 654.84375, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9901699960231781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009830006584525108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.64825744628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.20557556152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7510, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19982205480337142, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.02022706139832735, + "grad_norm": 0.14860881865024567, + "kept_token_count": 654.721875, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9897084474563599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01029155869036913, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5441589355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5817047119141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7520, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2002219572663307, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.01672776723280549, + "grad_norm": 0.1356402337551117, + "kept_token_count": 654.23125, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.990679794549942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009320213831961155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.4629699707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7302307128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7530, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.1993762508034706, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.015781895350664855, + "grad_norm": 0.10326801985502243, + "kept_token_count": 655.3703125, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9907301604747772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009269833285361528, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.8064361572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3740997314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7540, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.19984858483076096, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.016026577353477477, + "grad_norm": 0.12983667850494385, + "kept_token_count": 655.1046875, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9906269669532776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009373031370341778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.62403259277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2362548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7550, + "valid_token_count": 818.71875 + }, + { + "actual_token_mask_ratio": 0.20102912038564683, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.02008512634783983, + "grad_norm": 0.14184843003749847, + "kept_token_count": 653.7328125, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9906470954418183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009352898132055998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.7472686767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8291259765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7560, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2010035291314125, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.015288614947348834, + "grad_norm": 0.11989457905292511, + "kept_token_count": 653.9390625, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9909093499183654, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00909064020961523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1572692871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.50574645996096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7570, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19978588819503784, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.015552271343767642, + "grad_norm": 0.12220287322998047, + "kept_token_count": 654.3828125, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9909216105937958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00907838363200426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.2131011962891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4606201171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7580, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.20001226663589478, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.016262578126043082, + "grad_norm": 0.09890931844711304, + "kept_token_count": 654.715625, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.991038715839386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008961276430636645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.32330932617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.52549743652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7590, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.2003739982843399, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.01736007984727621, + "grad_norm": 0.12182068824768066, + "kept_token_count": 654.3015625, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9909673213958741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009032675717025995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.9054901123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.40331115722654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7600, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20024657547473906, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.015522310789674521, + "grad_norm": 0.107860267162323, + "kept_token_count": 654.4890625, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9908065497875214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009193454403430223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.4406433105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.37855834960936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7610, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.19980743080377578, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.01593206776306033, + "grad_norm": 0.0897713229060173, + "kept_token_count": 654.8453125, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9916993081569672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008300704136490821, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9027557373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.09897766113284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7620, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.199938502907753, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.017225325386971236, + "grad_norm": 0.11976249516010284, + "kept_token_count": 654.7359375, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9906747162342071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009325283486396075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.082080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.23448181152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7630, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19978941529989241, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.017627810034900902, + "grad_norm": 0.11928442120552063, + "kept_token_count": 655.134375, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9910701870918274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008929820824414491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.106103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.14336853027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7640, + "valid_token_count": 818.7 + }, + { + "actual_token_mask_ratio": 0.2004549339413643, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.01708280574530363, + "grad_norm": 0.12165424972772598, + "kept_token_count": 654.0921875, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9905361056327819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009463877882808447, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.8815399169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4203735351563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7650, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.1999243900179863, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.020757107343524695, + "grad_norm": 0.13670490682125092, + "kept_token_count": 654.7765625, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9906699776649475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00933000510558486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.88979187011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.8548217773438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7660, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19979955404996871, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.017413016874343157, + "grad_norm": 0.17102815210819244, + "kept_token_count": 654.9, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.991057276725769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008942726068198681, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.4208221435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7749328613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7670, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2003279820084572, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016913444362580778, + "grad_norm": 0.1266823709011078, + "kept_token_count": 654.4296875, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9916708171367645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008329187240451574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.1965087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.65166015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7680, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20099349766969682, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.015342137217521668, + "grad_norm": 0.1433325856924057, + "kept_token_count": 653.7953125, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9910667836666107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008933215495198965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.726953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5694305419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20027446001768112, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.020914411917328835, + "grad_norm": 0.11171474307775497, + "kept_token_count": 654.265625, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9900051534175873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009994847327470779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.84013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5603363037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7700, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2000138580799103, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.017801961582154035, + "grad_norm": 0.12110753357410431, + "kept_token_count": 654.459375, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9904764235019684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009523602202534676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.0475769042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1335052490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7710, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19991257339715957, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.015511038899421691, + "grad_norm": 0.12602739036083221, + "kept_token_count": 654.771875, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9906848430633545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009315156750380993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.42467346191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1646392822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7720, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20046326220035554, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.018510658480226993, + "grad_norm": 0.12455447763204575, + "kept_token_count": 654.1921875, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912348866462708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087651033885777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.8289428710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2812072753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7730, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19937691688537598, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.016053856071084737, + "grad_norm": 0.14114628732204437, + "kept_token_count": 655.4703125, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9911972463130951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008802756667137146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.6980010986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7604125976562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7740, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.19913387894630433, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.01683521466329694, + "grad_norm": 0.11199948191642761, + "kept_token_count": 655.4515625, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9909416854381561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009058299474418163, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.46654052734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.13376770019534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7750, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20008038282394408, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.016784920170903207, + "grad_norm": 0.11711706966161728, + "kept_token_count": 654.484375, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9902771770954132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009722824767231941, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.21885986328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1288177490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7760, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19900407791137695, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.02203482547774911, + "grad_norm": 0.1313544660806656, + "kept_token_count": 655.6421875, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0266, + "mgd_cos_sim": 0.9909174323081971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00908257281407714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4430389404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4150085449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7770, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.19976506680250167, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.016384859476238488, + "grad_norm": 0.10672635585069656, + "kept_token_count": 654.9625, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9903034567832947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009696543496102095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.21556701660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.88245544433596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19932861775159835, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.015768115315586327, + "grad_norm": 0.12396414577960968, + "kept_token_count": 655.0375, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.990885066986084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00911493431776762, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.1799041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9237487792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7790, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.19958814084529877, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.015109222289174795, + "grad_norm": 0.10145419836044312, + "kept_token_count": 654.8484375, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9911239683628082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008876031264662743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.39182739257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.62795104980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7800, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.20015216022729873, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.017344016022980212, + "grad_norm": 0.10984635353088379, + "kept_token_count": 654.6, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9896088063716888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010391193255782128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0923248291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.91514892578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7810, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20071838796138763, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.014374205935746431, + "grad_norm": 0.11615859717130661, + "kept_token_count": 654.1328125, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9912681460380555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008731843810528517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.68597717285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.56463012695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7820, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.1994588553905487, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.018656748719513415, + "grad_norm": 0.12176961451768875, + "kept_token_count": 655.1359375, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9906706631183624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00932933995500207, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.62279357910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1917724609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7830, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19985734969377517, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.016241124272346495, + "grad_norm": 0.11864309757947922, + "kept_token_count": 655.0265625, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913897275924682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008610262256115674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.0794189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3758117675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7840, + "valid_token_count": 818.646875 + }, + { + "actual_token_mask_ratio": 0.20013924837112426, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.0162291057407856, + "grad_norm": 0.15164707601070404, + "kept_token_count": 654.478125, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9906621932983398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009337796829640865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.3326110839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.04242553710935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7850, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.1996464103460312, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.017433628626167774, + "grad_norm": 0.10526186227798462, + "kept_token_count": 654.8921875, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9914007484912872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008599253743886948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4003570556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48729553222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.19976979345083237, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.015745341312140226, + "grad_norm": 0.1324063390493393, + "kept_token_count": 654.6859375, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9906760334968567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009323976840823889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.767724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.40257263183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7870, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20053717344999314, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.017194243054836987, + "grad_norm": 0.11378372460603714, + "kept_token_count": 654.0796875, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9913627922534942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00863720579072833, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.6390869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.40027770996096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7880, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.20003029704093933, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.017864912562072278, + "grad_norm": 0.11624724417924881, + "kept_token_count": 654.428125, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9907181143760682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009281877335160971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.28125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.4496765136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7890, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.2003363385796547, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.017877478897571564, + "grad_norm": 0.15019433200359344, + "kept_token_count": 654.2109375, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9899143755435944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.01008562119677663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.9138214111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.78777465820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7900, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.2004515841603279, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.017410069052129985, + "grad_norm": 0.1071290597319603, + "kept_token_count": 654.246875, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9915493607521058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008450624812394381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.244970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.585498046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7910, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19969335347414016, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.016260023321956397, + "grad_norm": 0.08805522322654724, + "kept_token_count": 654.8078125, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9916772067546844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00832279697060585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.4881652832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.61749267578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7920, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.19949057102203369, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.017402680311352014, + "grad_norm": 0.1069645881652832, + "kept_token_count": 655.0921875, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9910538792610168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008946118969470262, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.50885009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.23272094726565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7930, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.20012783706188203, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.01751660807058215, + "grad_norm": 0.13814637064933777, + "kept_token_count": 654.3140625, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9908454179763794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009154585748910904, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.790625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7534515380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7940, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.19957048892974855, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.017817715741693975, + "grad_norm": 0.14271314442157745, + "kept_token_count": 654.9375, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.991091400384903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008908612001687289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.57228698730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.70554504394534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7950, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2000921204686165, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.01911763222888112, + "grad_norm": 0.11136084794998169, + "kept_token_count": 654.4671875, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9915399789810181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008460021112114191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.13201599121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.82456665039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7960, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2008323684334755, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.019558376260101797, + "grad_norm": 0.11138201504945755, + "kept_token_count": 653.8875, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.99078568816185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009214313980191947, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.2373291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.61468505859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20131774991750717, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.01681690625846386, + "grad_norm": 0.12415999174118042, + "kept_token_count": 653.3859375, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9913511574268341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008648847788572311, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.2575439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2969451904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7980, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20001499205827714, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.01841426333412528, + "grad_norm": 0.1159796193242073, + "kept_token_count": 654.640625, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9904126048088073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009587385784834623, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.1534912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6266693115234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 7990, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.20027589350938796, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.017081717494875193, + "grad_norm": 0.14036265015602112, + "kept_token_count": 654.3921875, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9900173723697663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00998262595385313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.23908081054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.34746704101565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8000, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.199850195646286, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.014869376737624406, + "grad_norm": 0.1693030744791031, + "kept_token_count": 654.7625, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9910929441452027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008907042816281319, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.43359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7788024902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8010, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.19949223697185517, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.015894384868443013, + "grad_norm": 0.12319295853376389, + "kept_token_count": 655.1125, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9915204226970673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008479584939777851, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.25140991210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.82884521484374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8020, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19916778206825256, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.017045970913022755, + "grad_norm": 0.11410978436470032, + "kept_token_count": 655.4125, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.991083163022995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008916841447353363, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.77110290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.40020141601565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8030, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.19969750344753265, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.017955729831010105, + "grad_norm": 0.12112267315387726, + "kept_token_count": 655.1765625, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9909495711326599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009050427936017513, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8280975341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0618438720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8040, + "valid_token_count": 818.665625 + }, + { + "actual_token_mask_ratio": 0.2001111999154091, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.017624872736632824, + "grad_norm": 0.1298339068889618, + "kept_token_count": 654.615625, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9905439555644989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009456043411046267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.8119750976563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5417022705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8050, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2007487639784813, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.01741847237572074, + "grad_norm": 0.13829198479652405, + "kept_token_count": 654.053125, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9908692896366119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009130707662552594, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.7379455566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2680633544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8060, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2010202407836914, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.016275254357606173, + "grad_norm": 0.12658192217350006, + "kept_token_count": 653.6453125, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9911194026470185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00888059539720416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.8355987548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.21190185546874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8070, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19994991272687912, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.018298257421702147, + "grad_norm": 0.12146642059087753, + "kept_token_count": 654.740625, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9913552403450012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008644765056669713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.44179382324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.49290771484374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8080, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.1996823653578758, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.015855583362281322, + "grad_norm": 0.12217146903276443, + "kept_token_count": 654.88125, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9914083302021026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00859164847061038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.74779052734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5554138183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8090, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.1999991849064827, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.01992176044732332, + "grad_norm": 0.10612266510725021, + "kept_token_count": 654.5078125, + "learning_rate": 2.620342168475869e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9898156702518464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.010184340830892324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.01930541992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6977233886719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8100, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.20050636380910875, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.016104162484407426, + "grad_norm": 0.14424119889736176, + "kept_token_count": 654.3265625, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.991011506319046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008988510072231292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.02184143066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7824005126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8110, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.19964497685432434, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.018816543743014336, + "grad_norm": 0.1376183032989502, + "kept_token_count": 654.5953125, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.990786224603653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009213773906230927, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.05662841796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9062133789063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8120, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.19913075417280196, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.018111521285027264, + "grad_norm": 0.1315729022026062, + "kept_token_count": 655.2015625, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9913284003734588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00867161424830556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.35874328613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6744873046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8130, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20000452697277069, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.015164028760045767, + "grad_norm": 0.1009654849767685, + "kept_token_count": 654.5078125, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9909808754920959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009019116777926684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.527734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9805969238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8140, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.20091629028320312, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.017964307218790054, + "grad_norm": 0.128780797123909, + "kept_token_count": 653.828125, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9916660487651825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008333944715559483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.9671905517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5420623779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8150, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19961972236633302, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.01812892034649849, + "grad_norm": 0.1184457540512085, + "kept_token_count": 655.028125, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9908556044101715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009144399128854275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.23480529785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.8140106201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8160, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.19994429498910904, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.01641686148941517, + "grad_norm": 0.09109523892402649, + "kept_token_count": 654.909375, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9913091003894806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00869090436026454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.035888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.51109619140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8170, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.19974095076322557, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.015399952977895736, + "grad_norm": 0.10285001248121262, + "kept_token_count": 654.4984375, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9915718019008637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008428200613707304, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.2176788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.70201416015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8180, + "valid_token_count": 817.8578125 + }, + { + "actual_token_mask_ratio": 0.19934875071048735, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.019413707219064234, + "grad_norm": 0.10395637899637222, + "kept_token_count": 655.1640625, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.990407657623291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009592336043715477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.01580200195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.20274047851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8190, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20014602839946746, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.016689719911664725, + "grad_norm": 0.10320127755403519, + "kept_token_count": 654.434375, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.99093257188797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009067431464791298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.82940979003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5022857666016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.2000571995973587, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.017024780064821242, + "grad_norm": 0.12947428226470947, + "kept_token_count": 654.4421875, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9905711591243744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009428847581148148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.5671447753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.41339721679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8210, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.19972028583288193, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.016166295763105153, + "grad_norm": 0.10660082846879959, + "kept_token_count": 654.89375, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9912630617618561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873692212626338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9097045898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.785546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20013841837644578, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.01646442310884595, + "grad_norm": 0.13353563845157623, + "kept_token_count": 654.671875, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9910091876983642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008990813791751862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.21721801757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2150360107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8230, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20058012902736663, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.016670828498899937, + "grad_norm": 0.11378467082977295, + "kept_token_count": 654.2328125, + "learning_rate": 2.604816905526948e-05, + "loss": 0.021, + "mgd_cos_sim": 0.991313886642456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008686106745153665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.05401916503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1818481445313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8240, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.20047007650136947, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.017925636656582356, + "grad_norm": 0.1251285970211029, + "kept_token_count": 654.3171875, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9914109110832214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00858910521492362, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.04676208496096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1758605957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8250, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.19883735477924347, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.016574028134346008, + "grad_norm": 0.15935103595256805, + "kept_token_count": 655.2046875, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9906980097293854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009301999025046826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.1330139160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.85536804199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8260, + "valid_token_count": 817.8125 + }, + { + "actual_token_mask_ratio": 0.19991390854120256, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.01644097315147519, + "grad_norm": 0.11026111990213394, + "kept_token_count": 654.8046875, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9915931046009063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008406895212829113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.792724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.36919555664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8270, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.199666365981102, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.018579935748130084, + "grad_norm": 0.07932141423225403, + "kept_token_count": 655.01875, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9911124110221863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008887593541294337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.13445434570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4177551269531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8280, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.1998096987605095, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.018215054273605348, + "grad_norm": 0.17126668989658356, + "kept_token_count": 655.084375, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9905922651290894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009407737199217081, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.3222686767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.25266418457034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8290, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.2008561298251152, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.016803368367254733, + "grad_norm": 0.12977343797683716, + "kept_token_count": 653.9453125, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9911571264266967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008842870779335498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9038970947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.89515380859376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8300, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.19963976591825486, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.018578419648110866, + "grad_norm": 0.12699571251869202, + "kept_token_count": 654.9984375, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9913263201713562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008673673495650291, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.3579467773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.60169677734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8310, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.20051344633102416, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.016902972478419542, + "grad_norm": 0.1301620602607727, + "kept_token_count": 654.134375, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9901735782623291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009826407209038735, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.09386291503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.34801330566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8320, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20068739652633666, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.017455007322132586, + "grad_norm": 0.11773280054330826, + "kept_token_count": 653.7875, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912149310112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008785064332187176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.2739776611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.61438903808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8330, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.20091018527746202, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.01666141888126731, + "grad_norm": 0.1176399290561676, + "kept_token_count": 653.7625, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.99061821103096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009381779097020626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.76234741210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.85060729980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8340, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.20000145882368087, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.019286209251731634, + "grad_norm": 0.15168309211730957, + "kept_token_count": 654.4734375, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9903010249137878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009698982164263726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.827783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3432586669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8350, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.20072359293699266, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.01659839106723666, + "grad_norm": 0.13807350397109985, + "kept_token_count": 654.053125, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9912603676319123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008739639073610306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.56767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5398193359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8360, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19930775016546248, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.014602622669190168, + "grad_norm": 0.08833059668540955, + "kept_token_count": 654.934375, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9909575462341309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009042447805404663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.99840087890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0053009033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8370, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.20014722645282745, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.01696917423978448, + "grad_norm": 0.11483919620513916, + "kept_token_count": 654.4703125, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9909890770912171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009010921698063612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.02186279296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.48902587890626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8380, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19874168783426285, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.01665881471708417, + "grad_norm": 0.0956365168094635, + "kept_token_count": 655.5265625, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9914878249168396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008512165863066912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.14842834472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4649169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8390, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20032665133476257, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.01608030255883932, + "grad_norm": 0.1034567579627037, + "kept_token_count": 654.2734375, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9916207492351532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00837924787774682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9448577880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2183471679688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8400, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.1999335542321205, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.0180876842699945, + "grad_norm": 0.11144400388002396, + "kept_token_count": 654.6703125, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915506303310394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008449365850538015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.6643127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1072967529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8410, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19971799701452256, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.016980983316898346, + "grad_norm": 0.14732931554317474, + "kept_token_count": 654.790625, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9911218762397767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008878132794052362, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.95794982910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8080322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8420, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.19991810768842697, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.015339240711182357, + "grad_norm": 0.10506927222013474, + "kept_token_count": 654.83125, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9910827696323394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008917239028960466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.62151489257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4966278076172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8430, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.19936746805906297, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.01689055897295475, + "grad_norm": 0.12339746952056885, + "kept_token_count": 655.175, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9912928998470306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008707100246101617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.4534881591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.69769287109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8440, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20130119919776918, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.020025212410837413, + "grad_norm": 0.12326992303133011, + "kept_token_count": 653.775, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9914604902267456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0085395029745996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.10614318847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.11988830566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8450, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19968027621507645, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.016786008421331645, + "grad_norm": 0.12518718838691711, + "kept_token_count": 654.9296875, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9907796740531921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009220330230891704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9723297119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8460, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20069838762283326, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.017683727387338876, + "grad_norm": 0.12938469648361206, + "kept_token_count": 653.7625, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9912414491176605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008758565317839384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.73585205078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.70506591796874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8470, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.20063209384679795, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.015854956302791833, + "grad_norm": 0.12014719843864441, + "kept_token_count": 654.1640625, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9913380980491638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008661893662065268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.57664184570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1463958740234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8480, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20017368197441102, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.017020546458661558, + "grad_norm": 0.1281047761440277, + "kept_token_count": 654.29375, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.991496992111206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850300518795848, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8646575927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.26731872558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8490, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.20082456916570662, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.018043970875442027, + "grad_norm": 0.13904258608818054, + "kept_token_count": 653.8, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9906195044517517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009380503930151463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.0609100341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0658325195312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8500, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20009570866823195, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.016236817929893732, + "grad_norm": 0.1319325715303421, + "kept_token_count": 654.453125, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913916766643525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00860832342877984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1260650634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9991729736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8510, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20053202658891678, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.016050102282315494, + "grad_norm": 0.11932466924190521, + "kept_token_count": 654.25625, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9909752130508422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009024780802428722, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8011993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.06455993652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8520, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19969500601291656, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.01687350021675229, + "grad_norm": 0.15024834871292114, + "kept_token_count": 654.965625, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9912979900836945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008702010568231344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8946807861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2126403808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8530, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.20070452690124513, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.01801519822329283, + "grad_norm": 0.1327662616968155, + "kept_token_count": 653.8015625, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9914476037025451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008552389219403267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.48240966796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2254699707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8540, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.2000708818435669, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.014906729757785796, + "grad_norm": 0.12429448962211609, + "kept_token_count": 654.703125, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9915607213973999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008439290896058082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.8270965576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0103363037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8550, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20035519152879716, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.018944754265248775, + "grad_norm": 0.14356541633605957, + "kept_token_count": 654.2625, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9909127235412598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009087273571640252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.09409484863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1754974365234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8560, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.19996469616889953, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.020790432952344416, + "grad_norm": 0.09289976954460144, + "kept_token_count": 654.7265625, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9908023238182068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009197666123509408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9367736816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.46904907226565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8570, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.1999564751982689, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.01782748894765973, + "grad_norm": 0.1430376172065735, + "kept_token_count": 654.828125, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9910907924175263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008909219037741422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.6638458251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2645538330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8580, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.20088366717100142, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.01651705037802458, + "grad_norm": 0.16619645059108734, + "kept_token_count": 653.9515625, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9912481069564819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00875189397484064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.39788513183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.37816467285154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19983699768781663, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.015563232731074095, + "grad_norm": 0.14359650015830994, + "kept_token_count": 654.940625, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9909167885780334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009083206299692392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.77829895019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.80773010253904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8600, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20036290735006332, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.018156427703797818, + "grad_norm": 0.12589576840400696, + "kept_token_count": 654.59375, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9908832311630249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009116774238646031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.5597778320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8310577392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8610, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.2000058636069298, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.016716022323817015, + "grad_norm": 0.10395041853189468, + "kept_token_count": 654.6125, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9904907524585724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009509259648621083, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.03505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.39458618164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8620, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2006803497672081, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.018062417674809695, + "grad_norm": 0.12356353551149368, + "kept_token_count": 654.0765625, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915677666664123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008432240597903729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.48798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0104064941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8630, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.199728100001812, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.018156000506132842, + "grad_norm": 0.1400451362133026, + "kept_token_count": 654.8734375, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9910793006420135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008920697309076787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.85023193359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2405578613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8640, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19959665089845657, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01695410031825304, + "grad_norm": 0.10663779824972153, + "kept_token_count": 654.8515625, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9917895436286926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210478909313678, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9631622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7146881103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8650, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.19975219368934632, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.016078878892585635, + "grad_norm": 0.12791788578033447, + "kept_token_count": 654.775, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9915667057037354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008433285914361478, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.1792266845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.25164794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8660, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20000410974025726, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.018199227377772333, + "grad_norm": 0.12897989153862, + "kept_token_count": 654.4734375, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9910472154617309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008952783606946468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.6857482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8054626464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8670, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20038479268550874, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.018675254192203283, + "grad_norm": 0.14754782617092133, + "kept_token_count": 654.23125, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9907179057598114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009282093495130539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.40553283691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.53338012695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8680, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20004535466432571, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.015969729144126178, + "grad_norm": 0.11616938561201096, + "kept_token_count": 654.371875, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9909880340099335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009011949971318245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.9079162597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8168090820312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8690, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.20038269013166427, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.01490139365196228, + "grad_norm": 0.10562051087617874, + "kept_token_count": 654.3609375, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9912951648235321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008704830333590508, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9415588378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.68235473632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.20079414397478104, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.01895480640232563, + "grad_norm": 0.17831389605998993, + "kept_token_count": 653.84375, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.990937477350235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009062524046748876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.53867797851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0893310546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8710, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20009230226278304, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.016681342851370573, + "grad_norm": 0.13072249293327332, + "kept_token_count": 654.703125, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9907971978187561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009202803950756788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0226654052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1636657714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8720, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.20037150681018828, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.015896412078291178, + "grad_norm": 0.15026772022247314, + "kept_token_count": 654.6796875, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9912684798240662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008731512166559697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7801971435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7874237060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8730, + "valid_token_count": 818.7296875 + }, + { + "actual_token_mask_ratio": 0.20076377540826798, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.01657608086243272, + "grad_norm": 0.1253780573606491, + "kept_token_count": 654.0140625, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9915359258651734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008464077301323415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.39710693359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7670074462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8740, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19960274845361708, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.014741622377187014, + "grad_norm": 0.13443517684936523, + "kept_token_count": 654.8125, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9908094584941864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009190545044839382, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7259765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9819305419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8750, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.19973040223121644, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.015380808804184198, + "grad_norm": 0.10580725222826004, + "kept_token_count": 655.053125, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9915666103363037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0084333841688931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9889831542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8434661865234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8760, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.1998729705810547, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.01914870599284768, + "grad_norm": 0.1052745059132576, + "kept_token_count": 654.5703125, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9914315402507782, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00856844950467348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.1837615966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1927185058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8770, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.1995101436972618, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.01753139914944768, + "grad_norm": 0.14916464686393738, + "kept_token_count": 655.153125, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9913252651691437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008674740884453058, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7827667236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.55811767578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8780, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.1991860419511795, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.01673695454373956, + "grad_norm": 0.14039316773414612, + "kept_token_count": 655.2609375, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9908833026885986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009116692002862692, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4394165039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5516784667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8790, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.2001408591866493, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.013870255276560783, + "grad_norm": 0.13564464449882507, + "kept_token_count": 654.565625, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9911857545375824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008814227953553199, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2168731689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.38532104492185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19895006120204925, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.015529094915837049, + "grad_norm": 0.11060510575771332, + "kept_token_count": 655.5796875, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9906318306922912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009368174057453871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.6553131103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9729888916016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8810, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19991034865379334, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.017212420329451562, + "grad_norm": 0.12395612895488739, + "kept_token_count": 654.753125, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9913664877414703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008633514959365129, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.4767150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5494140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8820, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.19971214085817338, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.017891823034733533, + "grad_norm": 0.1121276468038559, + "kept_token_count": 654.9796875, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9910939157009124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008906068094074727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5463531494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3542205810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19935193806886672, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.019565766025334597, + "grad_norm": 0.10197403281927109, + "kept_token_count": 655.1421875, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9913288474082946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008671136852353812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.71009826660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2929168701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8840, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19876381307840346, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.018536678701639175, + "grad_norm": 0.10407890379428864, + "kept_token_count": 655.5890625, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912501454353333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008749856799840926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.664013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.18861999511716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8850, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19992597699165343, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.016989528201520443, + "grad_norm": 0.13176269829273224, + "kept_token_count": 654.696875, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9913561940193176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00864381082355976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.08864135742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1253997802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8860, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20032070875167846, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.01888400102034211, + "grad_norm": 0.1335756778717041, + "kept_token_count": 654.3140625, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9906832754611969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009316721651703119, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1429443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.35553283691405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8870, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20107343047857285, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.015673820301890372, + "grad_norm": 0.13215439021587372, + "kept_token_count": 653.8171875, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9910267412662506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008973288256675005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.70362854003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2216003417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19984490126371385, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.018098890129476786, + "grad_norm": 0.10019735991954803, + "kept_token_count": 654.8140625, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9912064075469971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00879359170794487, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.385888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.49265441894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8890, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20026479810476303, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.017219408601522445, + "grad_norm": 0.1021152213215828, + "kept_token_count": 654.41875, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9910658538341522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008934145420789718, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.66650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.70869140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8900, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20048182755708693, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.015561818610876799, + "grad_norm": 0.13356129825115204, + "kept_token_count": 654.328125, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9911533117294311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008846687525510788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4200836181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.05869750976564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.20111051052808762, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.017084290273487568, + "grad_norm": 0.11451496928930283, + "kept_token_count": 653.534375, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9909058749675751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009094114881008863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6368072509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.48834228515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8920, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.19948456585407257, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.021183341462165117, + "grad_norm": 0.11643343418836594, + "kept_token_count": 654.990625, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9907722234725952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009227792173624039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.50791625976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.80499267578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8930, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20088787674903869, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.018578880187124013, + "grad_norm": 0.1477031707763672, + "kept_token_count": 654.0640625, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9911181807518006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008881809748709201, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.15771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.78969421386716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8940, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.20002737492322922, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.016252302657812835, + "grad_norm": 0.10408179461956024, + "kept_token_count": 654.8359375, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9912123680114746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008787629939615727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8821075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0761322021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8950, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.19966541081666947, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.018664927035570145, + "grad_norm": 0.13423311710357666, + "kept_token_count": 654.78125, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9913928866386413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008607108891010285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4891021728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.6992431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8960, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19963785260915756, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.019633917976170777, + "grad_norm": 0.13950228691101074, + "kept_token_count": 654.7953125, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914028823375702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008597117476165294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1014343261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2253875732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8970, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19927823841571807, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.016951032914221287, + "grad_norm": 0.131342813372612, + "kept_token_count": 655.078125, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9911725699901581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008827438112348318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.20452880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3046569824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8980, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.1992918610572815, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.017561325803399086, + "grad_norm": 0.1118922233581543, + "kept_token_count": 655.275, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9914378225803375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008562187477946282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6160430908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.319580078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 8990, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19950582534074784, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.017340438347309826, + "grad_norm": 0.10611642152070999, + "kept_token_count": 654.9484375, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9913956046104431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00860439771786332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.68474731445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.634521484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9000, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2000188335776329, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.01847589062526822, + "grad_norm": 0.1574520766735077, + "kept_token_count": 654.8515625, + "learning_rate": 2.514827010117009e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9909810841083526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009018897078931332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7115203857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4454040527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9010, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.1999189242720604, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.018338813446462154, + "grad_norm": 0.15300793945789337, + "kept_token_count": 654.646875, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9908964335918427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00910358466207981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.44223937988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9906036376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9020, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20041221678256987, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.018716092128306626, + "grad_norm": 0.21110479533672333, + "kept_token_count": 654.3296875, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.990551096200943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009448913857340813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.5070068359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.36008911132814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9030, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.1998610645532608, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.01645566029474139, + "grad_norm": 0.11231380701065063, + "kept_token_count": 654.6921875, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.021, + "mgd_cos_sim": 0.990892106294632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00910789268091321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.6908905029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.80279846191405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9040, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19958483427762985, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.0156649268232286, + "grad_norm": 0.10577400773763657, + "kept_token_count": 655.253125, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9911021173000336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00889789303764701, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.5632019042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8265747070312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9050, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.20062652975320816, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.017312042601406574, + "grad_norm": 0.135574609041214, + "kept_token_count": 654.3421875, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9912421822547912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008757816907018424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8965179443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7579345703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9060, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20044585168361664, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.01812151921913028, + "grad_norm": 0.11601690202951431, + "kept_token_count": 654.24375, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9906406342983246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009359362721443176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.3986053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8515289306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9070, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.1994342491030693, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.016697790194302797, + "grad_norm": 0.11045762151479721, + "kept_token_count": 655.3734375, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9909577667713165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009042232763022184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5658477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7372161865234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9080, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.20038502812385559, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.01619683764874935, + "grad_norm": 0.138192281126976, + "kept_token_count": 654.4609375, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9911198198795319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00888017425313592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.66269836425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1682373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9090, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.20012753307819367, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.01741522392258048, + "grad_norm": 0.10235235840082169, + "kept_token_count": 654.5359375, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912101686000824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008789844810962677, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7406707763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.43457946777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9100, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.19943269193172455, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.015471205022186041, + "grad_norm": 0.13418003916740417, + "kept_token_count": 655.13125, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9906376361846924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009362371359020471, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.5448394775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5567687988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20028481930494307, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.017266342602670192, + "grad_norm": 0.11007176339626312, + "kept_token_count": 654.3671875, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9908622145652771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009137798380106688, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.3583221435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.38341064453124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9120, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20056262761354446, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.01611858857795596, + "grad_norm": 0.11099923402070999, + "kept_token_count": 654.2765625, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917989552021027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008201056439429522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.60576782226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.326904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9130, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.19989072531461716, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.01552510648034513, + "grad_norm": 0.09938602149486542, + "kept_token_count": 654.7984375, + "learning_rate": 2.49889445596515e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9910808145999909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008919195178896189, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.1711944580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2863372802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9140, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.19926297813653945, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.019479131419211627, + "grad_norm": 0.12440727651119232, + "kept_token_count": 655.2921875, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9907933115959168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00920668700709939, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.34825439453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7349548339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9150, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19977344423532487, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.015530061442404985, + "grad_norm": 0.10328823328018188, + "kept_token_count": 654.7578125, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9909119427204132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00908804526552558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.89739379882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.16510009765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9160, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19977508783340453, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.017774199414998294, + "grad_norm": 0.15276320278644562, + "kept_token_count": 654.771875, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.99085373878479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009146249014884233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.40444030761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.67600708007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9170, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.20074575692415236, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.016310086101293565, + "grad_norm": 0.09095537662506104, + "kept_token_count": 653.7703125, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9904301166534424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009569874126464128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.00432739257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5006866455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9180, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.19956334978342055, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.018597816675901414, + "grad_norm": 0.2051365226507187, + "kept_token_count": 655.1046875, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9909104883670807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009089517593383788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.45628356933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.80637817382814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9190, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.1999785527586937, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.017554168310016393, + "grad_norm": 0.1843922734260559, + "kept_token_count": 654.5578125, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9910129487514496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008987060841172933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.11085815429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.15039978027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19980860203504563, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.018566779792308807, + "grad_norm": 0.14767350256443024, + "kept_token_count": 654.628125, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9911582171916962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008841788303107024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.40472412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1295196533203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9210, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.20004730820655822, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.017716030683368446, + "grad_norm": 0.1584537923336029, + "kept_token_count": 654.5140625, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9904929399490356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009507066290825605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.44443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9775054931641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9220, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20018031895160676, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.017707750666886568, + "grad_norm": 0.12272625416517258, + "kept_token_count": 654.6234375, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9906294286251068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009370565228164197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.38401794433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6287384033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9230, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.19965571016073227, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.015658135432749985, + "grad_norm": 0.1319635510444641, + "kept_token_count": 654.9625, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9906636774539948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00933633567765355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2654632568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.70958251953124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9240, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20030163526535033, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.017737132590264083, + "grad_norm": 0.13044077157974243, + "kept_token_count": 654.4453125, + "learning_rate": 2.485252719338089e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9915075480937958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008492462430149316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.7330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6955322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9250, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.19923412799835205, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.018026583082973956, + "grad_norm": 0.13071104884147644, + "kept_token_count": 655.1578125, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915393233299256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008460680581629276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2941070556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.51246643066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9260, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20009839683771133, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.02049896037206054, + "grad_norm": 0.13934822380542755, + "kept_token_count": 654.9296875, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9911759853363037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008824026398360729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.9658142089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.53790283203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9270, + "valid_token_count": 818.765625 + }, + { + "actual_token_mask_ratio": 0.19985672384500502, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.018554060161113738, + "grad_norm": 0.1580810844898224, + "kept_token_count": 654.921875, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912972033023835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008702805172652006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0667388916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.30926818847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9280, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.19945316761732101, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.016137644834816455, + "grad_norm": 0.1368843913078308, + "kept_token_count": 654.978125, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917337834835053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008266211953014135, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.79115295410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.302587890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9290, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.1993931859731674, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.01803058050572872, + "grad_norm": 0.13031138479709625, + "kept_token_count": 655.0328125, + "learning_rate": 2.479003922205605e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9920077979564667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007992202416062355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.08660278320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5173065185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9300, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20093269646167755, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.016338490508496763, + "grad_norm": 0.1406269669532776, + "kept_token_count": 653.5828125, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9911914825439453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008808518666774035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.51476135253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.87627563476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9310, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.19995380640029908, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.01521870717406273, + "grad_norm": 0.1009899228811264, + "kept_token_count": 654.8375, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9909371614456177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009062838274985552, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3595306396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0102020263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9320, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20023393779993057, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.016643655858933924, + "grad_norm": 0.15601180493831635, + "kept_token_count": 654.4578125, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9909106433391571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009089349023997783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.8880981445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9319213867187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9330, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.19902988374233246, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.020899083092808725, + "grad_norm": 0.12772402167320251, + "kept_token_count": 655.465625, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9913336575031281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008666347153484822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.3648254394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.23793640136716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9340, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20008266419172288, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.017458479199558497, + "grad_norm": 0.11488090455532074, + "kept_token_count": 654.59375, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9910535335540771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008946466725319624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.8760986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.56430969238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20081360787153243, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.01641523102298379, + "grad_norm": 0.14843742549419403, + "kept_token_count": 654.0296875, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9913146674633027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008685337658971549, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2129150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5593017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.200247822701931, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.01529389787465334, + "grad_norm": 0.10068656504154205, + "kept_token_count": 654.4125, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9911525547504425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008847438544034959, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.98828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73370971679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19997743368148804, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014068979956209659, + "grad_norm": 0.09440122544765472, + "kept_token_count": 654.6078125, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.991479080915451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852092569693923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.68846740722654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2835754394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20057128965854645, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.016680837143212557, + "grad_norm": 0.14476153254508972, + "kept_token_count": 654.3078125, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.990914499759674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009085510671138764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4386779785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3222961425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9390, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.20014180690050126, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.017310500051826237, + "grad_norm": 0.11435005813837051, + "kept_token_count": 654.540625, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9910477578639985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008952239621430635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.76426086425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5751922607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9400, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2004803776741028, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.01810841402038932, + "grad_norm": 0.12289481610059738, + "kept_token_count": 654.4171875, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9908237159252167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009176285658031701, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1451416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.53537902832034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9410, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.2006349503993988, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015105507150292397, + "grad_norm": 0.1115359291434288, + "kept_token_count": 654.1328125, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9915778756141662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00842212736606598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.4596405029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.65321655273436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9420, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19930549412965776, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.015597914718091488, + "grad_norm": 0.11134467273950577, + "kept_token_count": 655.5453125, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9918550193309784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008144979178905488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1169189453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2982208251953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9430, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.19997582584619522, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.016201812773942947, + "grad_norm": 0.11256508529186249, + "kept_token_count": 654.653125, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9914478421211242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008552170637995005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.68829040527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.34522705078126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9440, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2000292271375656, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.0170343816280365, + "grad_norm": 0.16418202221393585, + "kept_token_count": 654.36875, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.991417670249939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00858232257887721, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.888037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.89117431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9450, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.19941065162420274, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.01782823819667101, + "grad_norm": 0.13923850655555725, + "kept_token_count": 655.0390625, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9905678153038024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009432198759168387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4931274414063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9968597412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9460, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20076982527971268, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.01766590280458331, + "grad_norm": 0.10465001314878464, + "kept_token_count": 654.196875, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9910043954849244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008995617553591729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.7595916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.82481384277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9470, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.2000524714589119, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.016530902218073608, + "grad_norm": 0.11914163827896118, + "kept_token_count": 654.478125, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9908334195613862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009166551753878593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.60707397460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73182983398436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9480, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.200158055126667, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.016277994867414235, + "grad_norm": 0.1044793501496315, + "kept_token_count": 654.3453125, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9914139926433563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008585993759334087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.079736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.37867126464846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9490, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.20010793060064316, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.01443723300471902, + "grad_norm": 0.11282606422901154, + "kept_token_count": 654.5, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9915308117866516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008469192683696747, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.14036560058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6317901611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9500, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2008521631360054, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.01668893499299884, + "grad_norm": 0.11895643919706345, + "kept_token_count": 653.990625, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9918159484863281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008184054493904113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3417694091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4596008300781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9510, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1990613803267479, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.01857575923204422, + "grad_norm": 0.14504766464233398, + "kept_token_count": 655.0296875, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9906367123126983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009363283216953278, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.54933776855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.16224365234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9520, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.19942080974578857, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.013748054578900337, + "grad_norm": 0.1065187081694603, + "kept_token_count": 655.0890625, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9911374032497406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008862580265849828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.08721923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7711547851562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9530, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.20041736960411072, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.016010372806340455, + "grad_norm": 0.0952690914273262, + "kept_token_count": 654.3625, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9917309701442718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008269038889557124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.03209228515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.16558837890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9540, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.20028023719787597, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.01592926373705268, + "grad_norm": 0.1160198524594307, + "kept_token_count": 654.4890625, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.991514366865158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008485628571361304, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.85926818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.75595703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9550, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19999948292970657, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.014202769752591849, + "grad_norm": 0.0986706018447876, + "kept_token_count": 654.7125, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9918821394443512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008117842208594083, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.358251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.27886962890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9560, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.1995047390460968, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.016752676479518412, + "grad_norm": 0.11647220700979233, + "kept_token_count": 654.83125, + "learning_rate": 2.444751372488677e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9915124654769898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00848754495382309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.68060302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.38955078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9570, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.19985754936933517, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.018215029500424862, + "grad_norm": 0.13172799348831177, + "kept_token_count": 655.0375, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9910218179225921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008978175558149814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.0212921142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1991424560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9580, + "valid_token_count": 818.64375 + }, + { + "actual_token_mask_ratio": 0.20001491159200668, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.01722291512414813, + "grad_norm": 0.1554294377565384, + "kept_token_count": 654.525, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9907993316650391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009200659580528736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.976123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8754913330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9590, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2002077654004097, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017268381360918284, + "grad_norm": 0.12026175856590271, + "kept_token_count": 654.56875, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9913422763347626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008657722361385822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.3524658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3287017822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9600, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20114130824804305, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.015323769487440585, + "grad_norm": 0.09408561885356903, + "kept_token_count": 653.509375, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9912772297859191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008722766954451799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.71495666503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.75400390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9610, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20010839998722077, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.01650079255923629, + "grad_norm": 0.10422175377607346, + "kept_token_count": 654.5046875, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.990870726108551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009129263926297426, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5781280517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9503479003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9620, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19950827658176423, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.016396039444953204, + "grad_norm": 0.10055974870920181, + "kept_token_count": 654.9953125, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9916277348995208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008372265659272671, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3536071777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9414367675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9630, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.1993241488933563, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.01633760193362832, + "grad_norm": 0.10081850737333298, + "kept_token_count": 655.071875, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9916877210140228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00831227945163846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.0292694091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7808807373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9640, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19970864951610565, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.016153109539300204, + "grad_norm": 0.11594544351100922, + "kept_token_count": 654.978125, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.991360092163086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008639889117330313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3998748779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.68140869140626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9650, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19991129785776138, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.01514968303963542, + "grad_norm": 0.12163829803466797, + "kept_token_count": 654.928125, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9916418731212616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008358128368854523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.7576538085938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3088897705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9660, + "valid_token_count": 818.575 + }, + { + "actual_token_mask_ratio": 0.2006387695670128, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.015558849088847637, + "grad_norm": 0.10326176136732101, + "kept_token_count": 654.1453125, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9917300164699554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008269977662712336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.88609619140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.62750549316405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20050283372402192, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015303526259958743, + "grad_norm": 0.09867151081562042, + "kept_token_count": 654.215625, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9918198227882385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008180176001042127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.0921936035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1078155517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9680, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20001433044672012, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.015528492350131273, + "grad_norm": 0.10684642195701599, + "kept_token_count": 654.4625, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9913424193859101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00865758527070284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.39317016601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.25807189941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9690, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2013162463903427, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.017374457512050866, + "grad_norm": 0.1139109656214714, + "kept_token_count": 653.5578125, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9911167323589325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008883274998515845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.03163757324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.50440979003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9700, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20013621896505357, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.016758596524596214, + "grad_norm": 0.11642130464315414, + "kept_token_count": 654.521875, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9911869168281555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008813072368502617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.4343688964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.388427734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9710, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2006722003221512, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.01682650959119201, + "grad_norm": 0.10167776793241501, + "kept_token_count": 654.2, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918371558189392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008162839524447917, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1106018066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3435546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9720, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.20038879215717315, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.014014453627169132, + "grad_norm": 0.10487330704927444, + "kept_token_count": 654.3359375, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9916289567947387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008371044509112835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6675964355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.03343811035154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9730, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.200502710044384, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.018685505166649817, + "grad_norm": 0.1921137124300003, + "kept_token_count": 654.0546875, + "learning_rate": 2.422753693407244e-05, + "loss": 0.023, + "mgd_cos_sim": 0.991350519657135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008649485558271408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4325714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2271026611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9740, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.2004622831940651, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.018084244756028055, + "grad_norm": 0.12975184619426727, + "kept_token_count": 654.2546875, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9914804339408875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008519573602825403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3938751220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4684112548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9750, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19986001551151275, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.016246150340884925, + "grad_norm": 0.12667325139045715, + "kept_token_count": 654.6, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9912066876888275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087933043949306, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.92467041015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3828643798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9760, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20012182891368865, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.015457818564027549, + "grad_norm": 0.08516255021095276, + "kept_token_count": 654.7046875, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9914486229419708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008551391772925854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.7267761230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9031646728516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9770, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.19947334378957748, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.018687007389962672, + "grad_norm": 0.1247982457280159, + "kept_token_count": 654.8359375, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9914171874523163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008582815900444985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0972595214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.737060546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9780, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.19996225386857985, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.018978744558990002, + "grad_norm": 0.10749708861112595, + "kept_token_count": 654.7421875, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9906111717224121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938883787021041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.78993530273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5377685546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9790, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.19959756731987, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.014932819176465273, + "grad_norm": 0.09794539958238602, + "kept_token_count": 655.1359375, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9916650652885437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008334947284311056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0109558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5080078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9800, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.1999105542898178, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.014565330324694514, + "grad_norm": 0.09661098569631577, + "kept_token_count": 654.4515625, + "learning_rate": 2.413601200572538e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9911538243293763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008846178557723761, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.2687591552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.151220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9810, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.19970899671316147, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.015220806188881397, + "grad_norm": 0.10486456751823425, + "kept_token_count": 654.846875, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9911061823368073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00889383303001523, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.50856628417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9223114013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9820, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.1999212920665741, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.01632701503112912, + "grad_norm": 0.15379950404167175, + "kept_token_count": 654.69375, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9912318229675293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008768185228109359, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.7023162841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9566650390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.19922979772090912, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.016776284109801053, + "grad_norm": 0.11538664251565933, + "kept_token_count": 655.075, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9910664498806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008933548629283906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8050231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.32542724609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9840, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.2000580221414566, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.017690489068627357, + "grad_norm": 0.15345792472362518, + "kept_token_count": 654.225, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9913956701755524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008604324050247669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66473999023435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9850, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.19936324208974837, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.01609900938346982, + "grad_norm": 0.11551209539175034, + "kept_token_count": 654.8953125, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9912155926227569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008784397225826978, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.3339019775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8413146972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9860, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.20037710219621657, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.016801497247070075, + "grad_norm": 0.11800576001405716, + "kept_token_count": 654.7359375, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9915301382541657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008469863049685954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.81011352539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.38880615234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9870, + "valid_token_count": 818.803125 + }, + { + "actual_token_mask_ratio": 0.19994298070669175, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.017017861641943455, + "grad_norm": 0.13224640488624573, + "kept_token_count": 654.659375, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9910721898078918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008927810937166214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.1863037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2880340576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9880, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20004684180021287, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.01722711306065321, + "grad_norm": 0.1497594267129898, + "kept_token_count": 654.6109375, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9908570766448974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009142910782247781, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.11809692382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.66205444335935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9890, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.20008863359689713, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.015987351816147567, + "grad_norm": 0.11917644739151001, + "kept_token_count": 654.4921875, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9913585364818573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008641453925520182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.62237243652345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.87314758300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9900, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.1996898651123047, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.016076279524713754, + "grad_norm": 0.1478625386953354, + "kept_token_count": 655.09375, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917426109313965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008257395867258311, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.86204833984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5821899414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9910, + "valid_token_count": 818.546875 + }, + { + "actual_token_mask_ratio": 0.20001392364501952, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.019330376852303745, + "grad_norm": 0.11427417397499084, + "kept_token_count": 654.4390625, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918384075164794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008161598164588214, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.16099853515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.89239807128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9920, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.20052978694438933, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.019137281645089387, + "grad_norm": 0.14949217438697815, + "kept_token_count": 654.2203125, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9916370451450348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008362971805036068, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.07964477539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.92425231933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9930, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20033275783061982, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.016244000662118195, + "grad_norm": 0.12060794979333878, + "kept_token_count": 654.478125, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9917158484458923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008284152671694755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2419708251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1172393798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9940, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.19907417446374892, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.016091719176620244, + "grad_norm": 0.1092456504702568, + "kept_token_count": 655.2265625, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9918132781982422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00818672962486744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2240692138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.47422485351564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20056588798761368, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01536488113924861, + "grad_norm": 0.08670362085103989, + "kept_token_count": 654.0796875, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9912191331386566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008780868630856276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.66363220214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.44085998535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9960, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19988590627908706, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.015721863880753518, + "grad_norm": 0.1258026659488678, + "kept_token_count": 654.9125, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9909267067909241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009073283430188895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1444427490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1232147216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9970, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.2006451055407524, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.016638074815273286, + "grad_norm": 0.11419019848108292, + "kept_token_count": 654.3515625, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.021, + "mgd_cos_sim": 0.991264146566391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873585231602192, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9219024658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3158325195312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9980, + "valid_token_count": 818.603125 + }, + { + "actual_token_mask_ratio": 0.19974257797002792, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.015371481142938136, + "grad_norm": 0.11012203991413116, + "kept_token_count": 654.83125, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9917877614498138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008212246559560299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3992218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6191833496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 9990, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20032116025686264, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.016394085343927146, + "grad_norm": 0.10436496883630753, + "kept_token_count": 654.2375, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9910728633403778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008927150350064039, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.51849670410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9776275634766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10000, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.199875707924366, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.014874870330095291, + "grad_norm": 0.12602268159389496, + "kept_token_count": 654.9203125, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9909072995185852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00909269079566002, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.2380706787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.83628540039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10010, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19958776384592056, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.01730847042053938, + "grad_norm": 0.1169077455997467, + "kept_token_count": 655.0109375, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.991543447971344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008456538151949643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9357025146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.56898498535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10020, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19948612600564958, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.017317884415388108, + "grad_norm": 0.12283749878406525, + "kept_token_count": 655.334375, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9909154295921325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009084563981741667, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8625549316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4101196289063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10030, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.19959504902362823, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.015767147205770014, + "grad_norm": 0.10889109969139099, + "kept_token_count": 654.8578125, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9920760691165924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007923912722617388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0420379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.94138488769534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.1997404530644417, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.014730853214859962, + "grad_norm": 0.12190509587526321, + "kept_token_count": 654.7125, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9915997743606567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008400222845375538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.8675018310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4706665039063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2003992274403572, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.015346537623554469, + "grad_norm": 0.12760670483112335, + "kept_token_count": 654.3546875, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9914701998233795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008529815264046193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.0818817138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.71412048339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10060, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.1992710515856743, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.015822204947471618, + "grad_norm": 0.14303945004940033, + "kept_token_count": 655.2640625, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9914903044700623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008509698510169982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.49036865234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.69724426269534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10070, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19952981621026994, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.01701510651037097, + "grad_norm": 0.11251717805862427, + "kept_token_count": 654.7921875, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9913962662220002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008603741694241763, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6266784667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3244201660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10080, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.19949734956026077, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.01585512375459075, + "grad_norm": 0.10182460397481918, + "kept_token_count": 654.959375, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9912226438522339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008777362294495106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3519592285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7368133544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10090, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.2001645162701607, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.0175110618583858, + "grad_norm": 0.11054540425539017, + "kept_token_count": 654.590625, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9919759869575501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008024005871266126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7386169433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3610076904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20043912082910537, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.01718605337664485, + "grad_norm": 0.13357385993003845, + "kept_token_count": 654.04375, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9909994721412658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009000521712005138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.9220672607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8669464111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10110, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.20002859085798264, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.01627142112702131, + "grad_norm": 0.1443062722682953, + "kept_token_count": 654.6671875, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9917186141014099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00828138729557395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.6416290283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.88087463378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10120, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19958143234252929, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.014806990325450898, + "grad_norm": 0.11884851008653641, + "kept_token_count": 654.7453125, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9919346392154693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008065372239798307, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.7891082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7916687011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10130, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.1996081739664078, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.016496868804097176, + "grad_norm": 0.14885392785072327, + "kept_token_count": 655.315625, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.991486644744873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00851336745545268, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.76666259765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0729461669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10140, + "valid_token_count": 818.7421875 + }, + { + "actual_token_mask_ratio": 0.20004960000514985, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.013891758862882853, + "grad_norm": 0.09901261329650879, + "kept_token_count": 654.609375, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9918540775775909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00814592493698001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6273559570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.300390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10150, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20063906908035278, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.015241467207670212, + "grad_norm": 0.1417725682258606, + "kept_token_count": 654.0390625, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9909379363059998, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009062053542584181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.194140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0964660644531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10160, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.19950264543294907, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.018815077282488345, + "grad_norm": 0.10744620114564896, + "kept_token_count": 655.1484375, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.991737574338913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008262419700622558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3369873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.03077087402346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10170, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19954377710819243, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.015758983325213194, + "grad_norm": 0.12001378834247589, + "kept_token_count": 654.971875, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9915038108825683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008496183063834905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.16949462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.55955810546874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10180, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20019768178462982, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.01375828180462122, + "grad_norm": 0.10210705548524857, + "kept_token_count": 654.428125, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9916355609893799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00836443230509758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.88111267089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3509094238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10190, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.20069993585348128, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.015289209131151438, + "grad_norm": 0.10740773379802704, + "kept_token_count": 654.2796875, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9910164058208466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008983594179153443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.137255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2365753173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10200, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20002056658267975, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.015702744014561176, + "grad_norm": 0.09981045126914978, + "kept_token_count": 654.675, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9915412843227387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008458717539906502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1443664550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.5821563720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10210, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.19965408593416215, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.017891623079776764, + "grad_norm": 0.12482596188783646, + "kept_token_count": 654.8703125, + "learning_rate": 2.35891932073586e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9918643534183502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008135648909956217, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8371307373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.90651550292966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10220, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19918859750032425, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.015398881584405898, + "grad_norm": 0.08345533162355423, + "kept_token_count": 655.059375, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9916139781475067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386021014302969, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3119323730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.06259765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10230, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.1995205909013748, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.015052540879696607, + "grad_norm": 0.09656936675310135, + "kept_token_count": 655.1125, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9913756310939789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008624373562633991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.66721496582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.51572265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2008146569132805, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.014551733620464802, + "grad_norm": 0.11378097534179688, + "kept_token_count": 653.8609375, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9914915800094605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850842883810401, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6506408691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.09413452148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10250, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20007923394441604, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.0168601063080132, + "grad_norm": 0.1002575159072876, + "kept_token_count": 654.50625, + "learning_rate": 2.353488759518022e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9917278051376343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008272193185985089, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.37778625488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.44208984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10260, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.1999543711543083, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.01683608116582036, + "grad_norm": 0.11972890794277191, + "kept_token_count": 654.6375, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9921701967716217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007829819526523351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.43475952148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83466186523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10270, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20079429298639298, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.017138138599693774, + "grad_norm": 0.11710995435714722, + "kept_token_count": 653.6890625, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9910727500915527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008927240595221519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.16335144042966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.34198608398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10280, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.20020462423563004, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.015583626925945282, + "grad_norm": 0.12319745868444443, + "kept_token_count": 654.3109375, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9919708609580994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008029148913919926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4609741210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9865753173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10290, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.19992676079273225, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.01682745749130845, + "grad_norm": 0.1125776395201683, + "kept_token_count": 654.728125, + "learning_rate": 2.348041605216242e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9916738927364349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00832610772922635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.4438446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.69801330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10300, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.1995253250002861, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.017241902742534877, + "grad_norm": 0.12977802753448486, + "kept_token_count": 654.8640625, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9909081876277923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009091814514249563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.03590087890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.11777954101564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10310, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.1999474585056305, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.014523704629391431, + "grad_norm": 0.11295446008443832, + "kept_token_count": 654.51875, + "learning_rate": 2.345311838740309e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9911038517951966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008896160684525967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.113037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5214813232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10320, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20054420381784438, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.01693922821432352, + "grad_norm": 0.1032906249165535, + "kept_token_count": 654.2875, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9914261877536774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008573803678154945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.51787719726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6137359619141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10330, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20067559629678727, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.01577748926356435, + "grad_norm": 0.14295825362205505, + "kept_token_count": 653.71875, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9906193852424622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009380606934428215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5292083740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3101837158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10340, + "valid_token_count": 817.84375 + }, + { + "actual_token_mask_ratio": 0.20064175873994827, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.016042072139680386, + "grad_norm": 0.09772956371307373, + "kept_token_count": 654.0296875, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910234451293946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008976545929908753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5791534423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3614044189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10350, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.20007925629615783, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.0170329668559134, + "grad_norm": 0.11380235850811005, + "kept_token_count": 654.7359375, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9917715013027191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008228519931435585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9566375732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.90648193359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10360, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.1988821178674698, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.015861183125525712, + "grad_norm": 0.09759540855884552, + "kept_token_count": 655.784375, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9919346451759339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008065339736640453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4971130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10370, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.20011315643787383, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.015453773457556963, + "grad_norm": 0.12349953502416611, + "kept_token_count": 654.5515625, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9910161674022675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008983844332396984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7723754882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2788604736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10380, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2006098985671997, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.01741653783246875, + "grad_norm": 0.11571680754423141, + "kept_token_count": 654.1171875, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9919570088386536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008043010998517276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 457.02418823242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.65997619628905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10390, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20129500329494476, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.016250304132699966, + "grad_norm": 0.10503873229026794, + "kept_token_count": 653.9015625, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9917056202888489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294380363076925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.8794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.83575744628905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10400, + "valid_token_count": 818.6984375 + }, + { + "actual_token_mask_ratio": 0.20098651945590973, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.016302220430225133, + "grad_norm": 0.11074822396039963, + "kept_token_count": 653.821875, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9913977265357972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008602286968380213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.36485595703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.46478881835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10410, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.1996611773967743, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.01618374167010188, + "grad_norm": 0.1451629251241684, + "kept_token_count": 654.7046875, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913526356220246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008647343702614307, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.57554931640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0427581787109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10420, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.20040736347436905, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014635116048157214, + "grad_norm": 0.09464170783758163, + "kept_token_count": 654.3125, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9915950715541839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008404934406280517, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.8268127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.59788208007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10430, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19935497194528579, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.01558681596070528, + "grad_norm": 0.1334538459777832, + "kept_token_count": 655.178125, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9915053367614746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008494667708873749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.37835998535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6161193847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10440, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2000524654984474, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.016410212870687245, + "grad_norm": 0.12364700436592102, + "kept_token_count": 654.4515625, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9914660215377807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008533980324864387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8190551757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5745056152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10450, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19912915378808976, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.017164597287774086, + "grad_norm": 0.10950565338134766, + "kept_token_count": 655.1453125, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9922222793102264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007777729537338018, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.87657165527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8735076904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10460, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.19999914914369582, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015432236064225436, + "grad_norm": 0.13242705166339874, + "kept_token_count": 654.625, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9919438779354095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008056114241480828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.10537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10470, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20067277699708938, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.01527548572048545, + "grad_norm": 0.11824905127286911, + "kept_token_count": 654.453125, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9913602352142334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00863976450636983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.99366149902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.91318969726564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10480, + "valid_token_count": 818.753125 + }, + { + "actual_token_mask_ratio": 0.20019352436065674, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.016598958801478148, + "grad_norm": 0.107452891767025, + "kept_token_count": 654.4015625, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9913967072963714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008603299502283335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.5210357666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7280639648437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20055281221866608, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.019332501292228698, + "grad_norm": 0.11536567658185959, + "kept_token_count": 654.1734375, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9913367748260498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008663216046988964, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1717132568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.22935485839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10500, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20019261687994003, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.01430016253143549, + "grad_norm": 0.09911756962537766, + "kept_token_count": 654.2578125, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9912911474704742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008708853274583817, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.34832153320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1076629638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10510, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.2005172923207283, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.015098183415830135, + "grad_norm": 0.12513650953769684, + "kept_token_count": 654.3, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9918177902698517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008182206563651562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0622192382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.99920959472655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10520, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.20086463689804077, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.01862455252557993, + "grad_norm": 0.14557407796382904, + "kept_token_count": 653.9390625, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9913856863975525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008614325243979693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5483123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9101806640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10530, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.199891360104084, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.016938608419150115, + "grad_norm": 0.1249644085764885, + "kept_token_count": 654.5625, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.991587883234024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008412102330476045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2586303710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0701965332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10540, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19965178966522218, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.017781918309628962, + "grad_norm": 0.13700881600379944, + "kept_token_count": 654.615625, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9910178661346436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008982146810740232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4223114013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2108581542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10550, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.20125630050897597, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.014975713286548852, + "grad_norm": 0.11193841695785522, + "kept_token_count": 653.44375, + "learning_rate": 2.312237989864752e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9919900476932526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008009946253150702, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.14052124023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.36407470703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10560, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19960569143295287, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.016136740706861018, + "grad_norm": 0.1100957989692688, + "kept_token_count": 655.1359375, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9915135502815247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008486454375088215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.09205322265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9700408935547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10570, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.19903386235237122, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.01458273669704795, + "grad_norm": 0.12180052697658539, + "kept_token_count": 655.1265625, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9916618943214417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008338102977722884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.4029998779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.46446533203124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10580, + "valid_token_count": 817.925 + }, + { + "actual_token_mask_ratio": 0.199358206987381, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.015947103779762984, + "grad_norm": 0.10931064188480377, + "kept_token_count": 655.146875, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9910626173019409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008937386330217123, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8324005126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.90537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10590, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19909556657075883, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.015010406635701657, + "grad_norm": 0.10803966969251633, + "kept_token_count": 655.575, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9913781523704529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008621847536414862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2871154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.81764831542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10600, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.19951466023921965, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.014333532098680735, + "grad_norm": 0.0991273745894432, + "kept_token_count": 655.3125, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.991935133934021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008064862992614507, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5601776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.37286071777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10610, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.19841976463794708, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.016968379728496075, + "grad_norm": 0.11618465930223465, + "kept_token_count": 655.9828125, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9918153405189514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00818466003984213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.61659545898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6499298095703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10620, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2007086232304573, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.01620536269620061, + "grad_norm": 0.10320435464382172, + "kept_token_count": 653.846875, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920040607452393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00799593860283494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.35390319824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.314111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10630, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2002953141927719, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.018347634095698596, + "grad_norm": 0.0971502810716629, + "kept_token_count": 654.43125, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9915077388286591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008492254000157119, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4995819091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0971221923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10640, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.20020972937345505, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.016486533172428607, + "grad_norm": 0.10609237849712372, + "kept_token_count": 654.509375, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.99202179312706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007978216093033553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.277099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2439819335938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10650, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20021455436944963, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.012583570554852486, + "grad_norm": 0.11769404262304306, + "kept_token_count": 654.3546875, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9917891502380372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210853766649961, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.6898681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.56455993652344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10660, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.20000624805688857, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.015349500067532062, + "grad_norm": 0.08563850820064545, + "kept_token_count": 654.4203125, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9915303111076355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008469682093709708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.22138061523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.82239990234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10670, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.20084849148988723, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.017355442233383655, + "grad_norm": 0.14244823157787323, + "kept_token_count": 653.825, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9911567091941833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008843300025910139, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.52666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65245666503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10680, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20139290988445283, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.016465052776038647, + "grad_norm": 0.11235040426254272, + "kept_token_count": 653.8375, + "learning_rate": 2.294084148900905e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9910108149051666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008989194314926863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5398193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4407135009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10690, + "valid_token_count": 818.728125 + }, + { + "actual_token_mask_ratio": 0.1992228925228119, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.016581427399069072, + "grad_norm": 0.10617676377296448, + "kept_token_count": 655.378125, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9917729914188385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008226995076984166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.37127990722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1578063964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10700, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.19966944307088852, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.01651537362486124, + "grad_norm": 0.11938109993934631, + "kept_token_count": 654.784375, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920278131961823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007972185686230659, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.56703186035156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10710, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.20054567456245423, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.01806253530085087, + "grad_norm": 0.1203707903623581, + "kept_token_count": 653.884375, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9915144920349122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008485507592558861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5407257080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.40660705566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10720, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.1993238866329193, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.016374328266829254, + "grad_norm": 0.1331280767917633, + "kept_token_count": 655.009375, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9917422473430634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008257729560136795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2094299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.31481018066404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10730, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.20015909671783447, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.01696805451065302, + "grad_norm": 0.11081468313932419, + "kept_token_count": 654.4171875, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9921468019485473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007853196747601033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.41217041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.65210876464846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10740, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.19922741055488585, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.016906555742025375, + "grad_norm": 0.12942031025886536, + "kept_token_count": 655.2890625, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9916652143001556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008334781043231487, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.7269348144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.57310485839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10750, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.1995750904083252, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01624941350892186, + "grad_norm": 0.10559778660535812, + "kept_token_count": 654.8421875, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9912817358970643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008718257118016482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4074737548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5699951171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10760, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.2009124994277954, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.015410533733665943, + "grad_norm": 0.10836759209632874, + "kept_token_count": 653.6078125, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9915755808353424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008424405939877033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0050354003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.11957092285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10770, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.19997449368238449, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.01333139818161726, + "grad_norm": 0.07488007843494415, + "kept_token_count": 654.8, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9917663514614106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008233647234737873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.09605102539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5097686767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10780, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.20015348196029664, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.014831780549138784, + "grad_norm": 0.148854598402977, + "kept_token_count": 654.434375, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9919704318046569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00802956884726882, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.9912139892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0093139648437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10790, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19942795038223265, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.017947463691234587, + "grad_norm": 0.13989707827568054, + "kept_token_count": 655.1421875, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9908016383647918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009198362194001674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3906951904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.76157531738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.1994437277317047, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.01573335751891136, + "grad_norm": 0.10535109788179398, + "kept_token_count": 655.2015625, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9920042634010315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00799575848504901, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2519287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.52849731445315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10810, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2005051478743553, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.015742534212768076, + "grad_norm": 0.11203622817993164, + "kept_token_count": 654.1640625, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9915470957756043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008452905807644129, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6696350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4741729736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10820, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20033922791481018, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.015007205586880446, + "grad_norm": 0.10009191185235977, + "kept_token_count": 654.60625, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9918158173561096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008184189070016145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.41754760742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8218933105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10830, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.20024960488080978, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.015307202748954297, + "grad_norm": 0.09957824647426605, + "kept_token_count": 654.5, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9912134170532226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008786592166870832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.2398223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5942443847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10840, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2000993847846985, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.01718682637438178, + "grad_norm": 0.11239996552467346, + "kept_token_count": 654.6734375, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.991963243484497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00803676201030612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.40137634277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.29296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10850, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.19897815734148025, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.015358984563499689, + "grad_norm": 0.11405596137046814, + "kept_token_count": 655.58125, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9917713642120362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0082286536693573, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4510498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3136413574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10860, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20058893859386445, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.016030331235378982, + "grad_norm": 0.10156115889549255, + "kept_token_count": 654.2140625, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9916203796863556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008379620872437954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.973291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3805877685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10870, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20019320845603944, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.015715664159506558, + "grad_norm": 0.11105924844741821, + "kept_token_count": 654.6796875, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9918916881084442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008108307607471943, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.21278686523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3809387207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10880, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.20054317265748978, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.014752689469605684, + "grad_norm": 0.10782518982887268, + "kept_token_count": 654.26875, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9917085886001586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008291414007544518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4287445068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4769622802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2005736231803894, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.01591263310983777, + "grad_norm": 0.10802250355482101, + "kept_token_count": 654.11875, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9908865094184875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009113489277660846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9223571777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.47312927246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10900, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.1997775599360466, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.015974719543009996, + "grad_norm": 0.1250017285346985, + "kept_token_count": 654.675, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918156564235687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008184352330863476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0094970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.03538818359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19962938576936723, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.01525412043556571, + "grad_norm": 0.12956956028938293, + "kept_token_count": 655.1203125, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9918270707130432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008172928914427757, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.3717895507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.48937072753904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10920, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19945717304944993, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.01647106958553195, + "grad_norm": 0.12108028680086136, + "kept_token_count": 655.0765625, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9911121904850007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008887817990034819, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9602386474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5997009277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10930, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20008659809827806, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.015467397309839726, + "grad_norm": 0.1346467286348343, + "kept_token_count": 654.8421875, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924526989459992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007547307014465332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.19167175292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.955859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10940, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.19942900389432908, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.014804027881473303, + "grad_norm": 0.12279795855283737, + "kept_token_count": 655.19375, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9912305235862732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008769478462636471, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0431182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6798858642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10950, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.198553729057312, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.014725452195852995, + "grad_norm": 0.10694559663534164, + "kept_token_count": 656.0390625, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9907731890678406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009226826950907707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8164031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.18214111328126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10960, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.19953646063804625, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.01853210860863328, + "grad_norm": 0.11092430353164673, + "kept_token_count": 654.9921875, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9916853368282318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008314668200910091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3936370849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0595001220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10970, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19984616190195084, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.01482719723135233, + "grad_norm": 0.10774563997983932, + "kept_token_count": 654.7671875, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9915512859821319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008448716159909964, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.66904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9576934814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.19973682463169098, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.017311489954590798, + "grad_norm": 0.11206675320863724, + "kept_token_count": 654.8703125, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9911300897598266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008869896363466977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.76732482910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.953076171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 10990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2002755030989647, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.015396277140825986, + "grad_norm": 0.11976729333400726, + "kept_token_count": 654.446875, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9919921815395355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008007824048399925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.2284851074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.59250793457034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11000, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19947572648525239, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.016584199201315642, + "grad_norm": 0.0888989046216011, + "kept_token_count": 654.9875, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9913574516773224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008642547577619553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.75750732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0383056640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11010, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.1993783161044121, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.01618965659290552, + "grad_norm": 0.10964807122945786, + "kept_token_count": 655.2890625, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917894840240479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210527244955301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.07997436523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4732238769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11020, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.19978374391794204, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.015772377047687768, + "grad_norm": 0.11211685836315155, + "kept_token_count": 654.7984375, + "learning_rate": 2.245843529059e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9913284957408905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008671498578041791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3277648925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.64541015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11030, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19940366894006728, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.014060413558036089, + "grad_norm": 0.11285858601331711, + "kept_token_count": 655.1109375, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9913943350315094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008605648018419742, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8526184082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3389862060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11040, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.19876208007335663, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.01600369531661272, + "grad_norm": 0.10968057811260223, + "kept_token_count": 655.625, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9914644658565521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008535525109618903, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6891815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3373657226563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11050, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.200061896443367, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.016482834797352554, + "grad_norm": 0.09686412662267685, + "kept_token_count": 654.8, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9914589822292328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008541014231741428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1764190673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8740631103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11060, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.19948670268058777, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.014347494952380658, + "grad_norm": 0.11804810166358948, + "kept_token_count": 654.959375, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.991779112815857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008220885787159205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.6686950683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.31927185058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11070, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.2001906678080559, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.016223144996911288, + "grad_norm": 0.13312925398349762, + "kept_token_count": 654.26875, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.992089468240738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007910528313368559, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11080, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.19970957338809966, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.014905746839940549, + "grad_norm": 0.11570217460393906, + "kept_token_count": 654.7640625, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9915839254856109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008416082616895438, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.01569213867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23341674804686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11090, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19895753115415574, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.015034673362970352, + "grad_norm": 0.09916860610246658, + "kept_token_count": 655.215625, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9918302536010742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008169746957719326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.86379089355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3698791503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11100, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.199660961329937, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.01564624225720763, + "grad_norm": 0.10582979768514633, + "kept_token_count": 654.8078125, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9915218114852905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008478178922086955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2207061767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0417999267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11110, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.1998081088066101, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.016411764640361072, + "grad_norm": 0.1323854923248291, + "kept_token_count": 654.7390625, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9920716881752014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007928316574543714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2744903564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.27984924316405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11120, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19928968101739883, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.017162521928548814, + "grad_norm": 0.10530966520309448, + "kept_token_count": 654.93125, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.990842080116272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009157924819737673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.87449340820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4535705566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11130, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.19928773045539855, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.014940638886764645, + "grad_norm": 0.09110770374536514, + "kept_token_count": 655.0578125, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9919870734214783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008012921176850795, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7057647705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4587036132813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11140, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.201114322245121, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.013491241354495287, + "grad_norm": 0.0884779840707779, + "kept_token_count": 653.696875, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9923501968383789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076498049311339855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9138549804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2528961181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11150, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20065720677375792, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.01724501708522439, + "grad_norm": 0.13717880845069885, + "kept_token_count": 654.05625, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920588791370392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007941118348389864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3824890136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4676574707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11160, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.1998508781194687, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.01665573874488473, + "grad_norm": 0.11510242521762848, + "kept_token_count": 654.4515625, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9918815910816192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008118397742509841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.79234619140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0315795898438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11170, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.20042628645896912, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.013996711000800133, + "grad_norm": 0.12164317071437836, + "kept_token_count": 654.225, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9915902018547058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008409802429378032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6383972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1288543701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11180, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.20016147345304489, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.01875227699056268, + "grad_norm": 0.09885642677545547, + "kept_token_count": 654.4859375, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9914558887481689, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008544108550995588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1895385742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.41785583496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11190, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20047089010477065, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.016751360427588225, + "grad_norm": 0.11040445417165756, + "kept_token_count": 654.16875, + "learning_rate": 2.22132610376864e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9914580047130584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854199780151248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.41172180175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0336639404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20038072615861893, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.017201739829033613, + "grad_norm": 0.09212774783372879, + "kept_token_count": 654.1234375, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9915824949741363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008417515829205514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.78632202148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9206298828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11210, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.2010473445057869, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.013993247039616108, + "grad_norm": 0.11334308236837387, + "kept_token_count": 653.771875, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9921443998813629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007855596486479044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.20933532714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5461730957031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11220, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2001261442899704, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.020104470755904914, + "grad_norm": 0.1352316290140152, + "kept_token_count": 654.1828125, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9910515964031219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008948395680636167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3313262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.37302856445314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11230, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.20031959116458892, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.015620404481887817, + "grad_norm": 0.11550404131412506, + "kept_token_count": 654.15625, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9904504776000976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009549520444124937, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.5810882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2766998291016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11240, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.20020903795957565, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.016450252570211887, + "grad_norm": 0.10973943769931793, + "kept_token_count": 654.1296875, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9913395881652832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008660408481955529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.15061950683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.42529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11250, + "valid_token_count": 817.875 + }, + { + "actual_token_mask_ratio": 0.19943357706069947, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.015440689120441674, + "grad_norm": 0.11426721513271332, + "kept_token_count": 655.1703125, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9915154576301575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008484536781907082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9293518066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7269653320312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11260, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.2007932633161545, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.014968979917466641, + "grad_norm": 0.10221405327320099, + "kept_token_count": 653.8890625, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9916119456291199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0083880671299994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9030426025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.01453857421876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11270, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1994343265891075, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.01397618455812335, + "grad_norm": 0.10672326385974884, + "kept_token_count": 655.3625, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9917796492576599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008220368064939976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6907043457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7679718017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11280, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.19938901513814927, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.013673006650060415, + "grad_norm": 0.12942297756671906, + "kept_token_count": 655.165625, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9920283436775208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007971661444753409, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.454736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.13841247558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11290, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19976218789815903, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.014850179757922887, + "grad_norm": 0.09129144251346588, + "kept_token_count": 654.9375, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9916664719581604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008333531487733126, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.4728485107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.32997131347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11300, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2000195026397705, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.01594899659976363, + "grad_norm": 0.11563403904438019, + "kept_token_count": 654.875, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.991248732805252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008751282934099436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.10455017089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.75892639160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11310, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.19970755726099015, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.01479733819141984, + "grad_norm": 0.09939530491828918, + "kept_token_count": 655.0171875, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.991844630241394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00815536705777049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.8263244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6891387939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11320, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.1998593732714653, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.015185021795332432, + "grad_norm": 0.09398162364959717, + "kept_token_count": 655.2265625, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9921506702899933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007849312666803599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.5255645751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9993408203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11330, + "valid_token_count": 818.890625 + }, + { + "actual_token_mask_ratio": 0.19953976273536683, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.017648428119719027, + "grad_norm": 0.10671595484018326, + "kept_token_count": 654.83125, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9907752752304078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009224731475114822, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.85205993652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0154174804687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11340, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.1999712958931923, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.016571190487593413, + "grad_norm": 0.12735460698604584, + "kept_token_count": 654.590625, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9916262090206146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008373792842030525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.4969482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6509155273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11350, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.19939850270748138, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.0166522397659719, + "grad_norm": 0.12941960990428925, + "kept_token_count": 655.0609375, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9907645463943482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009235454071313143, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.47914733886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6230499267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11360, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19949492812156677, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.016139496769756078, + "grad_norm": 0.08842094987630844, + "kept_token_count": 655.440625, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9916886627674103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008311336580663919, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2240783691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9093719482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11370, + "valid_token_count": 818.778125 + }, + { + "actual_token_mask_ratio": 0.19952143281698226, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.016146195679903032, + "grad_norm": 0.12060289084911346, + "kept_token_count": 655.146875, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9913785874843597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008621413353830576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.03489074707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.23480834960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11380, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.20003362596035004, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.017098371777683495, + "grad_norm": 0.11372629553079605, + "kept_token_count": 654.7359375, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9913230955600738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008676903415471315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.27972412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.85088500976565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2003390982747078, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.016566154453903435, + "grad_norm": 0.1170269027352333, + "kept_token_count": 654.421875, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920990526676178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00790095068514347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.07724914550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.31061401367185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11400, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.20018655210733413, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.01652960292994976, + "grad_norm": 0.11285825073719025, + "kept_token_count": 654.728125, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9912840902805329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008715895935893059, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4526031494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.12714233398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11410, + "valid_token_count": 818.6015625 + }, + { + "actual_token_mask_ratio": 0.20024862587451936, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.014221156109124422, + "grad_norm": 0.13490723073482513, + "kept_token_count": 654.3703125, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.991767555475235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008232442662119865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.0325531005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0604278564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11420, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19968695789575577, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.014406373631209135, + "grad_norm": 0.09800498932600021, + "kept_token_count": 654.6203125, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9921329200267792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00786708453670144, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.2014404296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.78265075683595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11430, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.2002831742167473, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.013611995335668325, + "grad_norm": 0.10641523450613022, + "kept_token_count": 654.58125, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992014592885971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007985406927764416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.82167358398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4436737060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11440, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19976480454206466, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.015461223386228084, + "grad_norm": 0.09351318329572678, + "kept_token_count": 654.6390625, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.991895979642868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008104030787944794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1256103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6270751953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11450, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.19964335709810258, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.01600452121347189, + "grad_norm": 0.14121286571025848, + "kept_token_count": 654.928125, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9910366237163544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008963368833065033, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.37534790039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.34784545898435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1997695669531822, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.013874799013137817, + "grad_norm": 0.10806025564670563, + "kept_token_count": 654.9390625, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9919214487075806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008078549802303315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.42918090820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8611236572266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11470, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.19960701167583467, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.015046530496329069, + "grad_norm": 0.10858549177646637, + "kept_token_count": 654.6328125, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9917021870613099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008297797106206418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.29693908691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.22365112304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11480, + "valid_token_count": 817.8890625 + }, + { + "actual_token_mask_ratio": 0.19939555823802949, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.015573932323604822, + "grad_norm": 0.12051388621330261, + "kept_token_count": 655.2015625, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921547949314118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007845205068588258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.96429138183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.77325134277345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11490, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20020836293697358, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.015284196659922599, + "grad_norm": 0.13245747983455658, + "kept_token_count": 654.409375, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9915941298007965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008405860047787428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.30783996582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8912414550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11500, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19967518895864486, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.016305947583168744, + "grad_norm": 0.1609678864479065, + "kept_token_count": 654.8828125, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9916130602359772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386946748942137, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0480590820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.71421508789064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11510, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20066119283437728, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.017776996176689864, + "grad_norm": 0.09822532534599304, + "kept_token_count": 654.2015625, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9912396907806397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008760302048176526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4511749267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3573974609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11520, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19962715357542038, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.018175489641726016, + "grad_norm": 0.11115259677171707, + "kept_token_count": 654.93125, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9920728206634521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007927175145596266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.92363891601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.94212036132814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11530, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19993217140436173, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.01747299749404192, + "grad_norm": 0.36994898319244385, + "kept_token_count": 654.7796875, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9918206989765167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008179300185292959, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.2975128173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8413879394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11540, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2001778781414032, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.017741051129996778, + "grad_norm": 0.1111087054014206, + "kept_token_count": 654.515625, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9916147708892822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00838521821424365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6771942138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.99668884277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11550, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20077979266643525, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.017086013033986092, + "grad_norm": 0.12831442058086395, + "kept_token_count": 654.2546875, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919927895069123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00800721049308777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2169250488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6656463623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11560, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.200312739610672, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.0151460912078619, + "grad_norm": 0.11473722010850906, + "kept_token_count": 654.39375, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9915622174739838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008437770418822765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6248016357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.07024841308595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11570, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.2003515988588333, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.017430408857762814, + "grad_norm": 0.08147489279508591, + "kept_token_count": 654.2453125, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9919477641582489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008052232302725315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6078277587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.35857543945315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11580, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20062683671712875, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.01766650890931487, + "grad_norm": 0.09113704413175583, + "kept_token_count": 653.9328125, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9922285377979279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007771472912281752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.8869934082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6776580810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11590, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20010802745819092, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.015574759058654308, + "grad_norm": 0.13122381269931793, + "kept_token_count": 654.696875, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.02, + "mgd_cos_sim": 0.991171681880951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008828309457749128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 454.2386749267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.32655639648436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11600, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.19962477684020996, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.01827489575371146, + "grad_norm": 0.12634684145450592, + "kept_token_count": 655.2140625, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.991200190782547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008799818437546491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.02509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.16249084472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11610, + "valid_token_count": 818.63125 + }, + { + "actual_token_mask_ratio": 0.19961818754673005, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.01916770776733756, + "grad_norm": 0.11059191077947617, + "kept_token_count": 654.9859375, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9907449007034301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009255120530724525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.47816162109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1622741699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11620, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20010679066181183, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.019739955943077803, + "grad_norm": 0.16026870906352997, + "kept_token_count": 654.55, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9915044784545899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008495518937706947, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.55957946777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.39280700683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19942313730716704, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.015718553122133015, + "grad_norm": 0.12166984379291534, + "kept_token_count": 654.98125, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9919882893562317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008011722564697265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.2148071289063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1888122558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11640, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.199860779941082, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.016201492212712765, + "grad_norm": 0.13524818420410156, + "kept_token_count": 654.896875, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9919628381729126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008037164434790611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8925750732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0378875732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11650, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20034312903881074, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.01485003950074315, + "grad_norm": 0.09866306930780411, + "kept_token_count": 654.371875, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.99219229221344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007807719614356756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.648291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.13059692382814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11660, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.19942911565303803, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.016538789961487055, + "grad_norm": 0.12473313510417938, + "kept_token_count": 655.0265625, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9916130542755127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386956434696913, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.50209350585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.937109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11670, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19981781840324403, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.014644274301826955, + "grad_norm": 0.11267015337944031, + "kept_token_count": 654.74375, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9914684474468232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008531548175960779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1206817626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1454162597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11680, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.20022722631692885, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.015421778056770563, + "grad_norm": 0.12932081520557404, + "kept_token_count": 654.4875, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9917945146560669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008205492328852416, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8777130126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9650482177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19979634881019592, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.018047480564564466, + "grad_norm": 0.1359008103609085, + "kept_token_count": 654.7265625, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9918197572231293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008180228620767593, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7080108642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.93267211914065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11700, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.20095294415950776, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.015693069342523812, + "grad_norm": 0.11428876966238022, + "kept_token_count": 653.6, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992001760005951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007998242415487767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8448516845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4473937988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11710, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.19995086044073104, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.014857433643192053, + "grad_norm": 0.11506704986095428, + "kept_token_count": 654.5671875, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9918320953845978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008167912531644107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3623931884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.16200866699216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11720, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.19998321682214737, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.01712136259302497, + "grad_norm": 0.13188444077968597, + "kept_token_count": 654.7578125, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9915235877037049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00847639637067914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.4548736572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2631317138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11730, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.20022716969251633, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.01648826990276575, + "grad_norm": 0.09412214159965515, + "kept_token_count": 654.4953125, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9916475057601929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0083524857647717, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1920593261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7040252685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11740, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19895003587007523, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.01800987608730793, + "grad_norm": 0.10918665677309036, + "kept_token_count": 655.4203125, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9918194651603699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008180535119026899, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5553863525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.157958984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11750, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.20010046660900116, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.016965916473418476, + "grad_norm": 0.11999497562646866, + "kept_token_count": 654.6515625, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9916713118553162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008328694198280574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.37529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35009765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11760, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.19999735653400422, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.015611767396330833, + "grad_norm": 0.12267222255468369, + "kept_token_count": 654.5828125, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9917776346206665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008222375251352786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2223724365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.338037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11770, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1994616374373436, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.013378201983869076, + "grad_norm": 0.11582168936729431, + "kept_token_count": 655.253125, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9922991871833802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007700805924832821, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1780487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8377655029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.20011380612850188, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.015970614086836577, + "grad_norm": 0.10259745270013809, + "kept_token_count": 654.5296875, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9913902044296264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008609799202531576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4198028564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9295684814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11790, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20038623213768006, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.01734781516715884, + "grad_norm": 0.13335373997688293, + "kept_token_count": 654.4359375, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913924694061279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008607526402920484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7186767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8212646484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11800, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.20021857023239137, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.01650896016508341, + "grad_norm": 0.1318315863609314, + "kept_token_count": 654.465625, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9921120703220367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007887934800237417, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.1728057861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.38978576660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11810, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20075433999300002, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.016641432233154774, + "grad_norm": 0.12529613077640533, + "kept_token_count": 653.9109375, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9916065514087677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008393448684364558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3824798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9615417480469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11820, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19978829622268676, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.016739862971007824, + "grad_norm": 0.10766635835170746, + "kept_token_count": 654.9078125, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9912199139595032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008780080359429121, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7408416748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8566009521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11830, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19998941719532012, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.01753462152555585, + "grad_norm": 0.1323498636484146, + "kept_token_count": 654.4390625, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.991401320695877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008598683308809996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.69956359863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2614501953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11840, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2002047434449196, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.016138555575162174, + "grad_norm": 0.12233906239271164, + "kept_token_count": 654.4578125, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9915344655513764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008465538453310727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4608978271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.62591552734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11850, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.1999792590737343, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.01502792676910758, + "grad_norm": 0.12479250878095627, + "kept_token_count": 654.6859375, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9917995572090149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008200419414788485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3199401855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9718933105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11860, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19946433007717132, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.016523841116577388, + "grad_norm": 0.09101103991270065, + "kept_token_count": 655.3015625, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922217845916748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077782184816896915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.15418395996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2104919433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11870, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.19953613132238388, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.0165089787915349, + "grad_norm": 0.1299876570701599, + "kept_token_count": 654.965625, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920925438404083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007907448895275592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8820556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36131896972654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11880, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.19981028139591217, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.016175026167184116, + "grad_norm": 0.1272912621498108, + "kept_token_count": 654.871875, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.991536819934845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00846318295225501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.00010070800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0070404052734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11890, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.20028752386569976, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.01563709191977978, + "grad_norm": 0.12330445647239685, + "kept_token_count": 654.490625, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9915112435817719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008488756138831377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.91010131835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.87049560546876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11900, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.19983382374048234, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.0167306549847126, + "grad_norm": 0.12612147629261017, + "kept_token_count": 654.71875, + "learning_rate": 2.116291843968583e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9914236605167389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00857635410502553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.52581176757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.14312133789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11910, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.19941234439611435, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.01636323845013976, + "grad_norm": 0.12298747897148132, + "kept_token_count": 655.1578125, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9912969589233398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008703054673969746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.1429168701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4807373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11920, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1994298830628395, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.018776098731905222, + "grad_norm": 0.11342571675777435, + "kept_token_count": 654.84375, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9920601904392242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007939811889082193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2975250244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.52276306152345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11930, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.2010212153196335, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.019075888302177192, + "grad_norm": 0.13459333777427673, + "kept_token_count": 653.784375, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9915926218032837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008407374657690526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3106719970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.04450073242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11940, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20077233463525773, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.015514963399618864, + "grad_norm": 0.0976412296295166, + "kept_token_count": 654.0828125, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9918993473052978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008100664801895618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.0696594238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2322204589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2004083886742592, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.018043712805956603, + "grad_norm": 0.13583430647850037, + "kept_token_count": 654.171875, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9916473269462586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008352678269147873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7628936767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6567047119141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11960, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.19947417378425597, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.015512006171047688, + "grad_norm": 0.12198356539011002, + "kept_token_count": 655.190625, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9917657494544982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008234243001788854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.20338134765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5394683837891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11970, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.20042289793491364, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.016200798097997904, + "grad_norm": 0.16771869361400604, + "kept_token_count": 654.5265625, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917768120765686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008223184663802386, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4096313476563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2091979980469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11980, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.1999294027686119, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.01418483778834343, + "grad_norm": 0.12896059453487396, + "kept_token_count": 654.5296875, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9919655799865723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008034419547766448, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.034375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5826873779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 11990, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20018749833106994, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.016449420526623727, + "grad_norm": 0.09972260892391205, + "kept_token_count": 654.5765625, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992016214132309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007983797136694193, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8286437988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.87323608398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12000, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20159414261579514, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.01797183249145746, + "grad_norm": 0.1375723034143448, + "kept_token_count": 653.2203125, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9919235050678253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00807649316266179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.72101440429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9271270751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12010, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2008182168006897, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.013650555163621902, + "grad_norm": 0.12833186984062195, + "kept_token_count": 653.94375, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9921502113342285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007849800772964955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.65150146484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.673486328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12020, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.1999586284160614, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.014723662473261356, + "grad_norm": 0.1069808080792427, + "kept_token_count": 654.3625, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.991981440782547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008018572255969048, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8457763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62648620605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12030, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.20021247118711472, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.015193476900458336, + "grad_norm": 0.09821650385856628, + "kept_token_count": 654.359375, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9923349916934967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00766499675810337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.4821472167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0243286132812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12040, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.1996980130672455, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.016468155942857266, + "grad_norm": 0.0969463363289833, + "kept_token_count": 654.9671875, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9921772003173828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007822804991155863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3732482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.59605407714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.1998790979385376, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.01784684378653765, + "grad_norm": 0.12269987910985947, + "kept_token_count": 654.859375, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9915978968143463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008402097597718238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5785888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6221984863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12060, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20019652396440507, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.014028010051697492, + "grad_norm": 0.09085793048143387, + "kept_token_count": 654.3078125, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9919378876686096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00806212006136775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.4110137939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.35248107910155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12070, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.19976281374692917, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.015099271945655347, + "grad_norm": 0.09147286415100098, + "kept_token_count": 654.984375, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9914370834827423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008562922105193138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.94000244140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2390441894531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12080, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.20065323412418365, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.013978640362620354, + "grad_norm": 0.0989941880106926, + "kept_token_count": 654.134375, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9915399253368378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00846007615327835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99993286132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7260711669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12090, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20006070733070375, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.016314608789980412, + "grad_norm": 0.08435387164354324, + "kept_token_count": 654.5859375, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9917508125305176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008249183278530836, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7571594238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1129913330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12100, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2007029801607132, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.016169415134936572, + "grad_norm": 0.1085410937666893, + "kept_token_count": 653.7703125, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9917223036289216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008277682214975357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.90538635253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.79913635253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12110, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.19955041706562043, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.015573574975132943, + "grad_norm": 0.12791401147842407, + "kept_token_count": 655.16875, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9916402637958527, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008359742909669876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.974365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8469665527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12120, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.19971411675214767, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.016272992920130493, + "grad_norm": 0.14240750670433044, + "kept_token_count": 654.9796875, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9911257863044739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00887420941144228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.89622802734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.96567687988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12130, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2000911146402359, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.015354849025607109, + "grad_norm": 0.11369604617357254, + "kept_token_count": 654.5875, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9913938462734222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00860614823177457, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8086334228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3007751464844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12140, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19935054928064347, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.01479273745790124, + "grad_norm": 0.0959005057811737, + "kept_token_count": 655.1609375, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9923344671726226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076655449345707895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1595916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.68263854980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12150, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.19972779005765914, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.016416839417070152, + "grad_norm": 0.11450988799333572, + "kept_token_count": 654.721875, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.992179811000824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007820186484605075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.21076049804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94916381835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12160, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20040688663721085, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.01468807589262724, + "grad_norm": 0.1082809716463089, + "kept_token_count": 654.08125, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9917949795722961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008205024059861898, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8708557128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0611267089844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12170, + "valid_token_count": 818.0171875 + }, + { + "actual_token_mask_ratio": 0.19947663247585296, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.017606884706765413, + "grad_norm": 0.11219531297683716, + "kept_token_count": 655.1109375, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920782446861267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00792175205424428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.965185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.56439514160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12180, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19994577318429946, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.014389093406498433, + "grad_norm": 0.15962152183055878, + "kept_token_count": 654.74375, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9910545587539673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008945439290255309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.9412109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6987274169922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12190, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.20023101270198823, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.01637623691931367, + "grad_norm": 0.10662411153316498, + "kept_token_count": 654.3640625, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9920850217342376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007914956752210855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.37091064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33283081054685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12200, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.19971071630716325, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.018330048862844707, + "grad_norm": 0.15704233944416046, + "kept_token_count": 654.7921875, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9919166505336762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00808334369212389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4246307373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7650482177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12210, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19992247968912125, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.014553467649966479, + "grad_norm": 0.12954333424568176, + "kept_token_count": 654.69375, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9921708524227142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007829139940440655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 455.7395446777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8769256591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12220, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.1996450111269951, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.016205326467752457, + "grad_norm": 0.0882934182882309, + "kept_token_count": 654.8765625, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9918139755725861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008186033833771944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3879058837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0604248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12230, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19903606176376343, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.016366902273148298, + "grad_norm": 0.12827074527740479, + "kept_token_count": 655.2953125, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9916704535484314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008329558372497558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.98951721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7847473144531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12240, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.1992425575852394, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.013898803945630788, + "grad_norm": 0.08423063904047012, + "kept_token_count": 655.2765625, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9923929750919342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007607020996510982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0318969726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2088104248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20032461285591124, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.016133719496428965, + "grad_norm": 0.12640970945358276, + "kept_token_count": 654.371875, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.990793639421463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00920634949579835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1678161621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8307281494141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12260, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2002012014389038, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.01798992557451129, + "grad_norm": 0.10183600336313248, + "kept_token_count": 654.728125, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9916199207305908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008380075637251138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.21844482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0032104492187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12270, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.19918299913406373, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.013978584576398135, + "grad_norm": 0.11308509111404419, + "kept_token_count": 655.603125, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9917266428470611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008273352868855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.50774536132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.296337890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12280, + "valid_token_count": 818.6703125 + }, + { + "actual_token_mask_ratio": 0.2002779021859169, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013565927557647229, + "grad_norm": 0.09830334037542343, + "kept_token_count": 654.6171875, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9919105648994446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008089439943432808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.13047180175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0992095947266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12290, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.20043296366930008, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.01816340135410428, + "grad_norm": 0.1417517513036728, + "kept_token_count": 654.4796875, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9921541571617126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007845851872116328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3286376953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07728576660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12300, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19988994747400285, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.01671643778681755, + "grad_norm": 0.1287965178489685, + "kept_token_count": 654.6890625, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9916867852210999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008313215058296918, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7718902587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.78271484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12310, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19955622851848603, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.015660551004111768, + "grad_norm": 0.12090383470058441, + "kept_token_count": 654.7515625, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9916848778724671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008315126318484545, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.96636352539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.64342346191404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12320, + "valid_token_count": 817.9859375 + }, + { + "actual_token_mask_ratio": 0.19891227036714554, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.014533668197691441, + "grad_norm": 0.1697910875082016, + "kept_token_count": 655.7953125, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9914444208145141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008555580954998731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.05402221679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.06182250976565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12330, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.19988182038068772, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.015797624737024306, + "grad_norm": 0.1114451065659523, + "kept_token_count": 654.590625, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9916154205799103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00838456628844142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2995086669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1237335205078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12340, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19970378130674363, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.016798178665339947, + "grad_norm": 0.09959093481302261, + "kept_token_count": 654.809375, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918851435184479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008114867191761733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.20782470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4066528320312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12350, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2005407527089119, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.015931677352637053, + "grad_norm": 0.11101523041725159, + "kept_token_count": 654.4921875, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9917318642139434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00826813094317913, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.174853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7912109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12360, + "valid_token_count": 818.671875 + }, + { + "actual_token_mask_ratio": 0.19916549921035767, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.014283058699220419, + "grad_norm": 0.10084889829158783, + "kept_token_count": 655.371875, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.991421502828598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00857849782332778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.23077392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5103271484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12370, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1998440593481064, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.01386874825693667, + "grad_norm": 0.0873357430100441, + "kept_token_count": 654.984375, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9919939279556275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008006072789430618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.99298400878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2919616699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12380, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20070662051439286, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.014949386846274138, + "grad_norm": 0.10764269530773163, + "kept_token_count": 654.0421875, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9920849144458771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007915076240897179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.15436096191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4403472900391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12390, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.20040348023176194, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.01485796170309186, + "grad_norm": 0.14676238596439362, + "kept_token_count": 654.6234375, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9919824719429016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008017528709024191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.3952392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9585754394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12400, + "valid_token_count": 818.690625 + }, + { + "actual_token_mask_ratio": 0.19969789385795594, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.0135410251095891, + "grad_norm": 0.11205461621284485, + "kept_token_count": 655.2203125, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.991965115070343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008034877292811871, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.85549011230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5376220703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12410, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.2000361680984497, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.014383675437420607, + "grad_norm": 0.11827734857797623, + "kept_token_count": 654.4734375, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924819946289063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007518008351325989, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.66578674316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8045257568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12420, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.20053900480270387, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.01826955173164606, + "grad_norm": 0.11413297057151794, + "kept_token_count": 654.0234375, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9915628254413604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008437187317758799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2646759033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1305419921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12430, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.20072322487831115, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.01365229655057192, + "grad_norm": 0.11213340610265732, + "kept_token_count": 654.3109375, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9923141598701477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007685839012265205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.68439331054685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.32847900390624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12440, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.19970863610506057, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.014331053011119366, + "grad_norm": 0.11587920039892197, + "kept_token_count": 654.996875, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9917526185512543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008247388154268264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.96783447265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6479217529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12450, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.20076369494199753, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.01328572304919362, + "grad_norm": 0.08351647108793259, + "kept_token_count": 654.1484375, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9920627534389496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007937251590192318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3485870361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.21368103027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12460, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.1994267836213112, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.01649169847369194, + "grad_norm": 0.1271543800830841, + "kept_token_count": 655.209375, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9917775213718414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008222471456974745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.52207336425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8686096191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12470, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.20046356767416001, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.013339726720005274, + "grad_norm": 0.10411177575588226, + "kept_token_count": 654.06875, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9919274032115937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008072594460099936, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.6186096191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.376123046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12480, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.19970015287399293, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016111658699810506, + "grad_norm": 0.08425518870353699, + "kept_token_count": 654.8984375, + "learning_rate": 2.02765928810275e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9921612679958344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007838731165975332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4592803955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04950256347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12490, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2004617765545845, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.014749891962856055, + "grad_norm": 0.12443547695875168, + "kept_token_count": 654.1625, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9916862428188324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008313760906457902, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.49359130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4086395263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12500, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19938447028398515, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.016592427995055913, + "grad_norm": 0.12664969265460968, + "kept_token_count": 655.353125, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9915709793567657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008429011702537537, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.49012756347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.28541259765626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12510, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.20031893849372864, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.012942017149180174, + "grad_norm": 0.09456117451190948, + "kept_token_count": 654.2875, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9917893946170807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00821061097085476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6011932373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4865325927734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12520, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20022408068180084, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.015843468299135566, + "grad_norm": 0.15010854601860046, + "kept_token_count": 654.553125, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9917380034923553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261996600776911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8373687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1944274902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12530, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19962628036737443, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.014059743843972683, + "grad_norm": 0.11596861481666565, + "kept_token_count": 654.9203125, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.991707718372345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008292287401854993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4039733886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3630065917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.19956860542297364, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.015386323351413012, + "grad_norm": 0.12138620764017105, + "kept_token_count": 654.8078125, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9916895151138305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008310498669743538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5826354980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.99231872558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12550, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.20064934939146042, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.017702332790941, + "grad_norm": 0.10580404847860336, + "kept_token_count": 653.903125, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9922700524330139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007729923911392688, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.20334777832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.325, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12560, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2004352927207947, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.015082448162138462, + "grad_norm": 0.1322929412126541, + "kept_token_count": 654.259375, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9918098390102387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008190174400806428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6708160400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8863037109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12570, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.19993463903665543, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.017314372584223747, + "grad_norm": 0.14298783242702484, + "kept_token_count": 654.7359375, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9926445007324218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073554983362555506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2099639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21100463867185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12580, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2003493681550026, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.015114070568233728, + "grad_norm": 0.11452259123325348, + "kept_token_count": 654.3875, + "learning_rate": 2.012149654706343e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9921860456466675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007813950907438994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.87281188964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.47693786621096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12590, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2007460117340088, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.013573984615504742, + "grad_norm": 0.11562837660312653, + "kept_token_count": 654.06875, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9921798467636108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007820158824324608, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.3299530029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.137646484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12600, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19945603460073472, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.01293229665607214, + "grad_norm": 0.0993523821234703, + "kept_token_count": 654.990625, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9920641243457794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007935870625078677, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6147064208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2287933349609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12610, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20008232593536376, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.016299835965037347, + "grad_norm": 0.12308581173419952, + "kept_token_count": 654.503125, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9914958953857422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008504112530499697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0694854736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7732757568359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12620, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.20008612424135208, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.016601389553397895, + "grad_norm": 0.10593444108963013, + "kept_token_count": 654.3703125, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9920304119586945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007969582825899124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.27971801757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.17730407714845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12630, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.19935162365436554, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.016733018029481173, + "grad_norm": 0.12133821099996567, + "kept_token_count": 654.9890625, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924467444419861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007553252018988132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1262908935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4478759765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12640, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.20065626502037048, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.014424215443432331, + "grad_norm": 0.09806118160486221, + "kept_token_count": 653.9328125, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9923670947551727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076328952796757225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5779693603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3366760253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12650, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.1995183616876602, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.015556923393160105, + "grad_norm": 0.09550782293081284, + "kept_token_count": 655.0046875, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9926192939281464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007380706258118153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.78749084472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4997802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.1989212766289711, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.015887079294770956, + "grad_norm": 0.09621123224496841, + "kept_token_count": 655.38125, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9920982003211976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007901796698570251, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4846923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65488891601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12670, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20051807165145874, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.014705717470496893, + "grad_norm": 0.08939031511545181, + "kept_token_count": 654.5015625, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922641456127167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007735839672386646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7161010742187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1171600341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12680, + "valid_token_count": 818.65625 + }, + { + "actual_token_mask_ratio": 0.2010403960943222, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.0165470611769706, + "grad_norm": 0.11043659597635269, + "kept_token_count": 654.059375, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922109723091126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007789020426571369, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.65916137695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.76645812988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12690, + "valid_token_count": 818.6359375 + }, + { + "actual_token_mask_ratio": 0.20092681646347046, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.014650618750602007, + "grad_norm": 0.11660332977771759, + "kept_token_count": 653.76875, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.992378157377243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007621842622756958, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.88134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9673309326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12700, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20076371133327484, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.01560618933290243, + "grad_norm": 0.09997968375682831, + "kept_token_count": 654.1, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922328352928161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007767171040177346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.12982177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9774597167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12710, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20124468058347703, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.017245178669691087, + "grad_norm": 0.10101030021905899, + "kept_token_count": 653.434375, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921064734458923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007893530651926994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.33310852050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9482879638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12720, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.20057404935359954, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.017614852637052536, + "grad_norm": 0.11351195722818375, + "kept_token_count": 654.0171875, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.992016875743866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007983114942908287, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.76934814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5073181152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12730, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20000273287296294, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.014331634901463986, + "grad_norm": 0.09015265852212906, + "kept_token_count": 654.63125, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925656735897064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007434307970106602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.4896942138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0077789306641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12740, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.1995370492339134, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.015379885770380497, + "grad_norm": 0.09101424366235733, + "kept_token_count": 654.8453125, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924343645572662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007565637025982142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0551452636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9621154785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12750, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.2005985900759697, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.016750779235735536, + "grad_norm": 0.08747446537017822, + "kept_token_count": 654.23125, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9919073402881622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008092665113508701, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4045715332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7723419189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12760, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19996651113033295, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.01754417335614562, + "grad_norm": 0.1164150983095169, + "kept_token_count": 654.6125, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9922030925750732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007796923723071813, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4927703857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7432159423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12770, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19903567135334016, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.015547950938344001, + "grad_norm": 0.11974948644638062, + "kept_token_count": 655.3515625, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9918650150299072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008134996239095926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.93564758300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1807098388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12780, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.19892241060733795, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.016570918541401625, + "grad_norm": 0.11319532245397568, + "kept_token_count": 655.346875, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9916772186756134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008322784490883351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.42938842773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7716491699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12790, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20039957612752915, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.015414456836879253, + "grad_norm": 0.10718535631895065, + "kept_token_count": 654.01875, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923889398574829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007611056696623564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5791412353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.12209167480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12800, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.20095207393169404, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.013869798462837934, + "grad_norm": 0.08721256256103516, + "kept_token_count": 653.875, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9921061336994171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007893861085176469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8920166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.95990295410155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12810, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19946849048137666, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.01577551607042551, + "grad_norm": 0.10101065784692764, + "kept_token_count": 655.2640625, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9921090245246887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007890977337956429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7989807128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7753662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12820, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.1992122083902359, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.015318733453750611, + "grad_norm": 0.1110672801733017, + "kept_token_count": 655.575, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9916974782943726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008302532322704791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.36443481445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.54808654785154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12830, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.20003135204315187, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015410943981260061, + "grad_norm": 0.10723965615034103, + "kept_token_count": 654.71875, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9912265539169312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008773447945713997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.47178344726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4304656982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12840, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.19970422089099885, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.01730423355475068, + "grad_norm": 0.1301974356174469, + "kept_token_count": 655.00625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9912351667881012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008764828648418188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2595916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.83628540039064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12850, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20029086768627166, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.015085327345877886, + "grad_norm": 0.09769966453313828, + "kept_token_count": 654.4546875, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9918958067893981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008104197215288877, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.6467803955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.93817443847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12860, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19883943349123, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.015645489282906055, + "grad_norm": 0.12683430314064026, + "kept_token_count": 655.45625, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.992147046327591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007852947525680064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1492553710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8205169677734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12870, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19884121417999268, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.014686869084835052, + "grad_norm": 0.10239941626787186, + "kept_token_count": 655.384375, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9916401147842407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008359875902533532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.20741577148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5028350830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12880, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.2000453069806099, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.016698648780584337, + "grad_norm": 0.14946238696575165, + "kept_token_count": 654.4453125, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9923106968402863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007689300645142794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4690734863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.60682983398436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12890, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20043976306915284, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.013486564997583628, + "grad_norm": 0.09311849623918533, + "kept_token_count": 654.2359375, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9923435926437378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007656396366655827, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9042144775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1669982910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.20084334015846253, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.01467804666608572, + "grad_norm": 0.12336869537830353, + "kept_token_count": 653.9359375, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9918932855129242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810670293867588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5935974121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0883361816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12910, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20012446194887162, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.015856998041272163, + "grad_norm": 0.1253758817911148, + "kept_token_count": 654.484375, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992330801486969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007669185940176249, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 463.8985595703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8420867919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12920, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.19895064681768418, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.01748589612543583, + "grad_norm": 0.12025383114814758, + "kept_token_count": 655.575, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9918501079082489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008149880263954402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6647186279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61053771972655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12930, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2006625860929489, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.01502908170223236, + "grad_norm": 0.12100624293088913, + "kept_token_count": 654.190625, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9918913125991822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810869289562106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3542419433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.13893127441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12940, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.1999681904911995, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.01655228706076741, + "grad_norm": 0.12231677770614624, + "kept_token_count": 654.875, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9915435492992402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008456455543637276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3397918701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.60968627929685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12950, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.20049704909324645, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.015753018762916328, + "grad_norm": 0.09931301325559616, + "kept_token_count": 654.1703125, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9915071189403534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008492875099182128, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7493133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3805236816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12960, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19893168807029724, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.014121198281645774, + "grad_norm": 0.128261998295784, + "kept_token_count": 655.2609375, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9918351233005523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008164871204644441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3352966308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.19940185546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12970, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.20011499226093293, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.01458120234310627, + "grad_norm": 0.11857976019382477, + "kept_token_count": 654.6234375, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9922852516174316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007714731805026531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.90960388183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0727508544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12980, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19904062002897263, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.01489606648683548, + "grad_norm": 0.13839267194271088, + "kept_token_count": 655.4390625, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9919668674468994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008033122774213553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5529724121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4342529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 12990, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19946853667497635, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.017356052063405514, + "grad_norm": 0.12270623445510864, + "kept_token_count": 654.990625, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9916269719600678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008373023755848408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.036669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.02036437988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13000, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20002844631671907, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.014823355525732041, + "grad_norm": 0.11398172378540039, + "kept_token_count": 654.6828125, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9919463157653808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008053692616522312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.94752197265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10987548828126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13010, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19931818693876266, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.015190189518034459, + "grad_norm": 0.1084766760468483, + "kept_token_count": 655.278125, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9916176497936249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008382351789623498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.8575714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5489562988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13020, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.199833482503891, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.014773362316191196, + "grad_norm": 0.12126129865646362, + "kept_token_count": 654.746875, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9922074258327485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077925904653966425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8374359130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.74896545410155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13030, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.2002454936504364, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.014087132643908262, + "grad_norm": 0.08456059545278549, + "kept_token_count": 654.3125, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9921951472759247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007804847881197929, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.67641296386716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.01770935058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13040, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19961010664701462, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.016699920780956744, + "grad_norm": 0.15886084735393524, + "kept_token_count": 655.0203125, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9922167062759399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077832980081439015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4332702636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.11004638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13050, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.1995835483074188, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.014365836698561906, + "grad_norm": 0.12537546455860138, + "kept_token_count": 654.91875, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924373388290405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007562669459730387, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.0790069580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57068786621096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13060, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19956681579351426, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.013156850263476372, + "grad_norm": 0.0976615622639656, + "kept_token_count": 655.1859375, + "learning_rate": 1.936870618531045e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9922235429286956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007776449900120497, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.8260833740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.13214721679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13070, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.20005430281162262, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.016647692210972308, + "grad_norm": 0.12885302305221558, + "kept_token_count": 654.6265625, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9919240474700928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008075963146984578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7564666748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2445465087891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13080, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19988246411085128, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.016696051694452763, + "grad_norm": 0.14173220098018646, + "kept_token_count": 655.1140625, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924211084842682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007578896265476942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2705474853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7202087402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13090, + "valid_token_count": 818.7703125 + }, + { + "actual_token_mask_ratio": 0.20084893256425856, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.01435611811466515, + "grad_norm": 0.14249759912490845, + "kept_token_count": 653.86875, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9921614706516266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007838521525263787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6072174072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9291290283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13100, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.19956209063529967, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.016939264815300704, + "grad_norm": 0.13074970245361328, + "kept_token_count": 654.85, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9919107139110566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008089307509362698, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.42027282714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8338928222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13110, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2006288528442383, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.013784242887049913, + "grad_norm": 0.10049702972173691, + "kept_token_count": 654.2109375, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9924845576286316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007515435945242644, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.570654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.90056762695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13120, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.20027168691158295, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.015608310233801603, + "grad_norm": 0.1133854016661644, + "kept_token_count": 654.5078125, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9916714429855347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008328546676784754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7723052978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6024475097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13130, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20054348558187485, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.013944986648857594, + "grad_norm": 0.10858508199453354, + "kept_token_count": 654.2328125, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9923303246498107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007669667806476354, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 455.29128112792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.955322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13140, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20030678063631058, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.015209108218550682, + "grad_norm": 0.10882516950368881, + "kept_token_count": 654.3265625, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9919205665588379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00807942533865571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.59975280761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.92124938964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13150, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20032815933227538, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.013801799714565277, + "grad_norm": 0.10974151641130447, + "kept_token_count": 654.4921875, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9921021699905396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007897836901247502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5870758056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1913604736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13160, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.20075736343860626, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.016151075158268212, + "grad_norm": 0.10103356093168259, + "kept_token_count": 654.1171875, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992050153017044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007949854992330075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.01796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.11488952636716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13170, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.1997503861784935, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.01482358118519187, + "grad_norm": 0.09410227090120316, + "kept_token_count": 654.6734375, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9918689906597138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008131001703441143, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7141815185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.99287109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13180, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.19990433305501937, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.017179704271256924, + "grad_norm": 0.11117391288280487, + "kept_token_count": 654.5859375, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9914476811885834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00855231462046504, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5385498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.619091796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13190, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.1997854605317116, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.013451474066823722, + "grad_norm": 0.12015122920274734, + "kept_token_count": 654.575, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9919256329536438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008074360154569148, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5255187988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2777954101563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13200, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.20059795528650284, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.015499819256365299, + "grad_norm": 0.09397531300783157, + "kept_token_count": 654.103125, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9922398269176483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007760182209312916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.307373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5429748535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13210, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19968048334121705, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.014803716354072094, + "grad_norm": 0.11482270061969757, + "kept_token_count": 654.5359375, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9918284058570862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008171592187136412, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7574737548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1786315917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13220, + "valid_token_count": 817.846875 + }, + { + "actual_token_mask_ratio": 0.19997057169675828, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.015532216615974903, + "grad_norm": 0.09820810705423355, + "kept_token_count": 654.696875, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925958395004273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074041646905243395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.48037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.75201416015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13230, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.199923600256443, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.013453343417495489, + "grad_norm": 0.09417923539876938, + "kept_token_count": 654.5078125, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9921416163444519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007858376856893301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1914001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3096923828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13240, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.19951801747083664, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.015184262674301863, + "grad_norm": 0.09484197199344635, + "kept_token_count": 655.034375, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9919286787509918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008071320597082376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4198516845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9730621337891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13250, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.1993713155388832, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.01729006115347147, + "grad_norm": 0.11946409195661545, + "kept_token_count": 655.415625, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923114001750946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007688596658408642, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.12377014160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9579376220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13260, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.2013121947646141, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.014428410632535815, + "grad_norm": 0.09180117398500443, + "kept_token_count": 653.0265625, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9925186991691589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00748129803687334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.4326538085937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4398681640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13270, + "valid_token_count": 817.621875 + }, + { + "actual_token_mask_ratio": 0.19939871579408647, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.01485622227191925, + "grad_norm": 0.10258769989013672, + "kept_token_count": 655.30625, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9924933671951294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007506630755960942, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.88689575195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9023834228516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13280, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.19993455111980438, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.015804239735007285, + "grad_norm": 0.1274566799402237, + "kept_token_count": 654.621875, + "learning_rate": 1.901946228439179e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9916555404663085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008344460278749466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8574279785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14552917480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13290, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20030492842197417, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.014218210522085429, + "grad_norm": 0.12453141808509827, + "kept_token_count": 654.2265625, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.992737454175949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007262542471289634, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6664245605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7342987060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13300, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19982402473688127, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.01503866259008646, + "grad_norm": 0.10903766751289368, + "kept_token_count": 654.5953125, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9912894010543823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008710600715130568, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.7301086425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4155700683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13310, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.2002372294664383, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.01855731038376689, + "grad_norm": 0.16603513062000275, + "kept_token_count": 654.6265625, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9920167326927185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007983259949833154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.89337768554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.43092041015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13320, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.20017070919275284, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.01617188863456249, + "grad_norm": 0.08897079527378082, + "kept_token_count": 654.49375, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920965433120728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007903456408530473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.76930236816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0968505859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13330, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20095746517181395, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.014671970065683127, + "grad_norm": 0.12883135676383972, + "kept_token_count": 653.746875, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9923528373241425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007647165097296238, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.33562927246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9779052734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13340, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.19933868795633317, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.014114117529243232, + "grad_norm": 0.12813037633895874, + "kept_token_count": 654.7921875, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9924897670745849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007510235160589218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6816711425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.90086059570314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13350, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.20064352601766586, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.017753726243972777, + "grad_norm": 0.1344919204711914, + "kept_token_count": 654.1890625, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9921447098255157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007855290547013283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.65337524414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9060852050781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13360, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.199139304459095, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.01257725628092885, + "grad_norm": 0.0949377492070198, + "kept_token_count": 655.31875, + "learning_rate": 1.889186760191666e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9930780410766602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006921981368213892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.0714569091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.40825805664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13370, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2002854824066162, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.014482884760946035, + "grad_norm": 0.12335334718227386, + "kept_token_count": 654.7765625, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992458701133728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007541299611330032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5096160888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9646057128906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13380, + "valid_token_count": 818.7625 + }, + { + "actual_token_mask_ratio": 0.19995662569999695, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.014843299519270658, + "grad_norm": 0.15416742861270905, + "kept_token_count": 654.803125, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.991873687505722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008126304112374782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1577453613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.36943359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13390, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.20032401382923126, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.015876880660653115, + "grad_norm": 0.18224529922008514, + "kept_token_count": 654.325, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.991920804977417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00807918505743146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.1643402099609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.24888916015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13400, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.19935432225465774, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.016962569765746595, + "grad_norm": 0.11672820895910263, + "kept_token_count": 655.2765625, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9915545046329498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008445483166724443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.52056274414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13410, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.19954053312540054, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.017221238277852536, + "grad_norm": 0.12660212814807892, + "kept_token_count": 655.1, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921211361885071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007878861110657453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.23662719726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5526885986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13420, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2006891891360283, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01694964598864317, + "grad_norm": 0.15293507277965546, + "kept_token_count": 654.3046875, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9928565979003906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007143396697938442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4697692871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.30291137695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13430, + "valid_token_count": 818.5859375 + }, + { + "actual_token_mask_ratio": 0.1990398034453392, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.01598702520132065, + "grad_norm": 0.1291089504957199, + "kept_token_count": 655.4296875, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923752129077912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007624785229563713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6562194824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9947570800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13440, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.1997793957591057, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.014676048513501882, + "grad_norm": 0.16714036464691162, + "kept_token_count": 654.8296875, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9921176612377167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007882344163954258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2311248779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.64275512695315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13450, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.19982936680316926, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.01430846583098173, + "grad_norm": 0.11254527419805527, + "kept_token_count": 654.609375, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924104928970336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075895044952631, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.82452697753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4770904541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13460, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19899795055389405, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.015963535849004985, + "grad_norm": 0.1360720694065094, + "kept_token_count": 655.3953125, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9915885984897613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008411403838545084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.64044799804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.517041015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13470, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19969253987073898, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.015595827717334032, + "grad_norm": 0.09358956664800644, + "kept_token_count": 654.934375, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922771513462066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007722841110080481, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.56361694335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.89190979003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19932015091180802, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.013849014975130558, + "grad_norm": 0.08373033255338669, + "kept_token_count": 655.0875, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9922370910644531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007762913405895233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1047088623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.69793090820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.20005431324243544, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.015829273406416176, + "grad_norm": 0.10133633017539978, + "kept_token_count": 654.45625, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9924112915992737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007588709704577923, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3129089355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.41065368652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13500, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20044275373220444, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.01734390975907445, + "grad_norm": 0.10919696092605591, + "kept_token_count": 654.10625, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9918895363807678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008110456075519324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6839874267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.13336486816405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13510, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.19976041913032533, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.01316364724189043, + "grad_norm": 0.13814465701580048, + "kept_token_count": 654.7546875, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9924365639686584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075634347274899484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3232879638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.13734741210936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13520, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.20041525214910508, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.01436052368953824, + "grad_norm": 0.09569782018661499, + "kept_token_count": 654.2796875, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9919374406337738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008062550704926253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7296600341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.358837890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20029258728027344, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.015200078394263983, + "grad_norm": 0.11761590093374252, + "kept_token_count": 654.11875, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9920864224433898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791358044371009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.58953247070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2590728759766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13540, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.20025646388530732, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.016105577861890197, + "grad_norm": 0.08457359671592712, + "kept_token_count": 654.4703125, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9925314068794251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007468604855239392, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.91114196777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0293670654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13550, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.1993655890226364, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.016817095596343277, + "grad_norm": 0.12269346415996552, + "kept_token_count": 655.2, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918209612369537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008179030288010836, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.40320739746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0763366699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13560, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2002093181014061, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.016561046056449414, + "grad_norm": 0.08399760723114014, + "kept_token_count": 654.4984375, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919789791107178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008021019119769334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0012725830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.64812622070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13570, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20012375265359877, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.015915140695869922, + "grad_norm": 0.10745348036289215, + "kept_token_count": 654.559375, + "learning_rate": 1.855551723485441e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9918308734893799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081691256724298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.03529663085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.59454650878905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13580, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.20023606717586517, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.017221377696841955, + "grad_norm": 0.10937321931123734, + "kept_token_count": 654.24375, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9918862164020539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008113779127597809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6338348388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.55517578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13590, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.19979758262634278, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.015548964124172926, + "grad_norm": 0.1320035755634308, + "kept_token_count": 654.6390625, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921259343624115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007874054741114378, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5799835205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4132080078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13600, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.19979171752929686, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.013043208327144385, + "grad_norm": 0.11714693903923035, + "kept_token_count": 654.9265625, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9914197146892547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00858028270304203, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.974658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2291961669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13610, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20025319755077362, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.01526435548439622, + "grad_norm": 0.12309819459915161, + "kept_token_count": 654.25625, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.991368156671524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00863184928894043, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.87938537597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.811376953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13620, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19980107992887497, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014180626533925534, + "grad_norm": 0.12602049112319946, + "kept_token_count": 655.0140625, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9922828435897827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007717141043394804, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.38662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3513488769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13630, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.20032420754432678, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.013707131426781417, + "grad_norm": 0.12254280596971512, + "kept_token_count": 654.484375, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9923352181911469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007664780132472515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.42405090332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.50672607421876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13640, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.2004740223288536, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.016336142551153898, + "grad_norm": 0.1332015097141266, + "kept_token_count": 654.315625, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9919708609580994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008029149658977986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3772827148438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5257202148438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13650, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20029909610748292, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.013578570168465375, + "grad_norm": 0.11555710434913635, + "kept_token_count": 654.4234375, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9925185203552246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007481483276933432, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.94375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2564971923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19850172549486161, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.015499948058277369, + "grad_norm": 0.12495388090610504, + "kept_token_count": 656.10625, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9916328251361847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00836716778576374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1620635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4389709472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13670, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.2002338781952858, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.0186981332488358, + "grad_norm": 0.16701385378837585, + "kept_token_count": 654.2953125, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918250381946564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008174956310540438, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2548461914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62706298828124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.20033917725086212, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.01583650978282094, + "grad_norm": 0.10255232453346252, + "kept_token_count": 654.35, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9912783801555634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008721613511443137, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.76121826171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7453094482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13690, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.20003995299339294, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.016530546918511392, + "grad_norm": 0.10657231509685516, + "kept_token_count": 654.4890625, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920429229736328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007957080844789743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.00921020507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23899536132814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13700, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.19983456134796143, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.016319177765399217, + "grad_norm": 0.15414129197597504, + "kept_token_count": 654.5921875, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9918892085552216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008110779616981745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4073486328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7316467285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.19954897165298463, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.014116086065769196, + "grad_norm": 0.11192122846841812, + "kept_token_count": 655.2796875, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9916268527507782, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008373140171170235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.87859497070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04682006835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13720, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.20059892684221267, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.0180102976039052, + "grad_norm": 0.11457838863134384, + "kept_token_count": 653.8515625, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9921516001224517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007848403323441744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9114959716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.34056396484374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13730, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.20050314366817473, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.01381820598617196, + "grad_norm": 0.10108181834220886, + "kept_token_count": 654.465625, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927111387252807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007288866583257914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0463806152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6225921630859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13740, + "valid_token_count": 818.59375 + }, + { + "actual_token_mask_ratio": 0.2007808193564415, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.01784644741564989, + "grad_norm": 0.13038115203380585, + "kept_token_count": 654.0578125, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.992681473493576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007318533957004547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.94451904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29847412109376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13750, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.20009724348783492, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.01657672617584467, + "grad_norm": 0.11383868753910065, + "kept_token_count": 654.496875, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9915647089481354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008435277733951807, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.35218811035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.39818115234374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.1997295454144478, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.012803190667182208, + "grad_norm": 0.10798340290784836, + "kept_token_count": 655.2421875, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9926189601421356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007381032034754753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.2983123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.07801818847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13770, + "valid_token_count": 818.7765625 + }, + { + "actual_token_mask_ratio": 0.20011291205883025, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.013489193189889193, + "grad_norm": 0.10079257935285568, + "kept_token_count": 654.6046875, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9921971738338471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007802837155759334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2782897949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.72430114746095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.1991899222135544, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.01390651185065508, + "grad_norm": 0.09181138128042221, + "kept_token_count": 655.0671875, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9923739433288574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007626069523394108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.4753082275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.73639526367185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13790, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.1995338261127472, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.013841607701033353, + "grad_norm": 0.0964168831706047, + "kept_token_count": 654.890625, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9921357870101929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007864215970039367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.3570983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.26014709472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19932316690683366, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.014154188521206378, + "grad_norm": 0.10293348878622055, + "kept_token_count": 655.3515625, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9919933259487153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008006663341075183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.70537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.80588684082034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13810, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19941047132015227, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.016040003020316364, + "grad_norm": 0.11548074334859848, + "kept_token_count": 654.978125, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917271852493286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008272820431739093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5026153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9768798828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13820, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.20087087601423265, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.01700351070612669, + "grad_norm": 0.12598170340061188, + "kept_token_count": 653.9953125, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9917891621589661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00821084026247263, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0123321533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.97997131347654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13830, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.19948284476995468, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.016906442865729333, + "grad_norm": 0.10786625742912292, + "kept_token_count": 655.1734375, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9920846462249756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007915363647043705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.41753540039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.82899475097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13840, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.1988935127854347, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.014952084794640542, + "grad_norm": 0.12371572852134705, + "kept_token_count": 655.353125, + "learning_rate": 1.812029268085074e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9919661462306977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008033854607492685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7862060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9623077392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13850, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.2001175731420517, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.012794384174048901, + "grad_norm": 0.11472868919372559, + "kept_token_count": 654.696875, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.992523604631424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074763943441212176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7930053710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1409149169922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13860, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19998819828033448, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.016304422169923782, + "grad_norm": 0.09692425280809402, + "kept_token_count": 654.765625, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9922776997089386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007722301874309778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.22132568359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5702697753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13870, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2002228692173958, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.01597465854138136, + "grad_norm": 0.10204509645700455, + "kept_token_count": 654.6140625, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923436164855957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007656386494636536, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 454.9779510498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.95567321777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13880, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.20016600638628007, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.015604164358228445, + "grad_norm": 0.10476483404636383, + "kept_token_count": 654.1984375, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9918901920318604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810981160029769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.1446136474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.86101684570315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13890, + "valid_token_count": 817.9140625 + }, + { + "actual_token_mask_ratio": 0.20004739761352539, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.014555560145527124, + "grad_norm": 0.10730420798063278, + "kept_token_count": 654.5546875, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.991929042339325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008070974238216877, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.95717163085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05540466308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13900, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19958401024341582, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.014873414672911167, + "grad_norm": 0.1250539869070053, + "kept_token_count": 654.8109375, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9930726110935211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069273743778467175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.74302062988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1813659667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20039601773023605, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.013690402545034886, + "grad_norm": 0.1215422973036766, + "kept_token_count": 654.1375, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.992301857471466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076981582678854465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.22267150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.42401428222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13920, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.20017848312854766, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.013196079805493355, + "grad_norm": 0.12649421393871307, + "kept_token_count": 654.28125, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.992919671535492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007080318685621023, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4897827148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3933502197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13930, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.19949414134025573, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.013325594644993544, + "grad_norm": 0.11472811549901962, + "kept_token_count": 655.2375, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9917683184146882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008231679722666741, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2815643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.37132568359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13940, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19966111928224564, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.015780533477663995, + "grad_norm": 0.10164519399404526, + "kept_token_count": 654.7765625, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9921236097812652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007876385282725096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.94256286621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5394989013672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13950, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20030554085969926, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.015086352545768022, + "grad_norm": 0.11433016508817673, + "kept_token_count": 654.4625, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927734792232513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007226512394845485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.6894500732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.50538024902346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13960, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.19983192533254623, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.017037081625312568, + "grad_norm": 0.10181009024381638, + "kept_token_count": 654.5265625, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9924489617347717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075510511174798015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9779479980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9341796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13970, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.2004008948802948, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.013598037324845791, + "grad_norm": 0.11431123316287994, + "kept_token_count": 653.978125, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926277577877045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007372233550995588, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.76633911132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3937652587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13980, + "valid_token_count": 817.8828125 + }, + { + "actual_token_mask_ratio": 0.20047454237937928, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.014828681573271751, + "grad_norm": 0.11218856275081635, + "kept_token_count": 654.015625, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9920108377933502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007989160902798175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3891174316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3096160888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 13990, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.20112811923027038, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.012838632892817258, + "grad_norm": 0.09794390946626663, + "kept_token_count": 653.7890625, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9918526947498322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008147304132580756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9149597167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8156433105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14000, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20014082491397858, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.015239578485488892, + "grad_norm": 0.13211944699287415, + "kept_token_count": 654.0546875, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922093391418457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007790656387805938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0760040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.50261840820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14010, + "valid_token_count": 817.7140625 + }, + { + "actual_token_mask_ratio": 0.1997906118631363, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.015937922056764364, + "grad_norm": 0.08337001502513885, + "kept_token_count": 654.7984375, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9924146175384522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007585376687347889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.97704162597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6283203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14020, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19940873980522156, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.013371384236961603, + "grad_norm": 0.06872490793466568, + "kept_token_count": 654.9828125, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9923234403133392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007676561456173658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1009460449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.92937622070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14030, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.2007473811507225, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.012568074464797973, + "grad_norm": 0.08815625309944153, + "kept_token_count": 653.9953125, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9924900710582734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007509943097829819, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.19593505859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6994232177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14040, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20037606358528137, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.01772568253800273, + "grad_norm": 0.11589445918798447, + "kept_token_count": 654.4234375, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9921694755554199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007830525655299425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.0008514404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.52199096679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14050, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20015396773815156, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.015203047264367341, + "grad_norm": 0.10547276586294174, + "kept_token_count": 654.36875, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9914775967597962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852240165695548, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.947314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2039367675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14060, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.1997997298836708, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.011780169140547515, + "grad_norm": 0.11621428281068802, + "kept_token_count": 655.2390625, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9920109450817108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007989043928682804, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.11488952636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6557556152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14070, + "valid_token_count": 818.8453125 + }, + { + "actual_token_mask_ratio": 0.19987836182117463, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.013710459880530834, + "grad_norm": 0.11300534009933472, + "kept_token_count": 654.6203125, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.991741555929184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008258446957916022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.42005920410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.92005920410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14080, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.20041994005441666, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.014239669684320689, + "grad_norm": 0.10091141611337662, + "kept_token_count": 653.959375, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9914948642253876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850513530895114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.27008056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4246520996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14090, + "valid_token_count": 817.878125 + }, + { + "actual_token_mask_ratio": 0.19975119978189468, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.014220460411161184, + "grad_norm": 0.10285164415836334, + "kept_token_count": 654.875, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9921423196792603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007857685443013906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.88433837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3207946777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14100, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19995813220739364, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.01434435616247356, + "grad_norm": 0.13841569423675537, + "kept_token_count": 654.3421875, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9923776686191559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007622328493744135, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.92583618164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5713592529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14110, + "valid_token_count": 817.8859375 + }, + { + "actual_token_mask_ratio": 0.1994781270623207, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.0159511006437242, + "grad_norm": 0.09651533514261246, + "kept_token_count": 655.0984375, + "learning_rate": 1.768230436366165e-05, + "loss": 0.02, + "mgd_cos_sim": 0.991992998123169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008007009141147136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2869354248047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5050384521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14120, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.1991260603070259, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.014765775389969349, + "grad_norm": 0.08757923543453217, + "kept_token_count": 655.3109375, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9925190329551696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007480950746685267, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.1990570068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15223083496096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14130, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2004876732826233, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.013163618091493846, + "grad_norm": 0.10219980031251907, + "kept_token_count": 654.2453125, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926860690116882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007313927263021469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.4338775634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5249847412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14140, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.20049886554479598, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.01274023549631238, + "grad_norm": 0.11853080242872238, + "kept_token_count": 654.5015625, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9925783753395081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007421631086617708, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.21715393066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9745330810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14150, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.20009151101112366, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.014862944558262824, + "grad_norm": 0.1073576956987381, + "kept_token_count": 654.496875, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.991566824913025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008433171734213828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0994110107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9014465332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14160, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.19914018958806992, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.014543329551815986, + "grad_norm": 0.1298135221004486, + "kept_token_count": 655.38125, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927075743675232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072924232110381125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.21178283691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.22711791992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.200436469912529, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.014700760040432215, + "grad_norm": 0.1338340938091278, + "kept_token_count": 654.2765625, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922027468681336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007797249779105187, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.38457946777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8663055419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14180, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20002991259098052, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.015497547015547753, + "grad_norm": 0.1297352910041809, + "kept_token_count": 654.5578125, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9919078171253204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00809218194335699, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4656677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3619140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2009356677532196, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.014941792842000723, + "grad_norm": 0.09371896088123322, + "kept_token_count": 653.7859375, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9923638522624969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076361545361578465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.52149963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.35602416992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14200, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20081206709146499, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.015722753666341304, + "grad_norm": 0.11399554461240768, + "kept_token_count": 654.084375, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9920226335525513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007977354992181062, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.66942443847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9306335449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14210, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.1993677705526352, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.016664947383105755, + "grad_norm": 0.09293106198310852, + "kept_token_count": 654.921875, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9922963321208954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007703674584627151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.30352783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.18040466308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14220, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.20089685916900635, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.015518462844192981, + "grad_norm": 0.15164129436016083, + "kept_token_count": 653.9578125, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9924942135810852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007505797687917948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.89032592773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.25828247070314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14230, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.19980360120534896, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.01436392143368721, + "grad_norm": 0.1307688057422638, + "kept_token_count": 654.4953125, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922936677932739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007706315536051988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1594696044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8948059082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14240, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.19990040510892867, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.016499038413167, + "grad_norm": 0.11299624294042587, + "kept_token_count": 654.4828125, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992163997888565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00783599866554141, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.17287292480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.24835815429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.2005890667438507, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.01413527736440301, + "grad_norm": 0.08710576593875885, + "kept_token_count": 654.23125, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9917055249214173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294466882944107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.77294006347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0607666015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14260, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2008109509944916, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.0157543140463531, + "grad_norm": 0.1290634274482727, + "kept_token_count": 653.884375, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9927383303642273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261674478650093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3311828613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9359069824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14270, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.1993955999612808, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.014308819454163313, + "grad_norm": 0.11979285627603531, + "kept_token_count": 655.3109375, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9915119409561157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008488053549081087, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.6658477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8448028564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14280, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.2008783996105194, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.014060365036129951, + "grad_norm": 0.0881185382604599, + "kept_token_count": 653.95625, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.992555570602417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007444441318511963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.44993286132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4994171142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.20048217177391053, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.01594186406582594, + "grad_norm": 0.08555901795625687, + "kept_token_count": 654.2171875, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9920561909675598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007943809777498246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7055419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8273498535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14300, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19985619485378264, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.018228751420974732, + "grad_norm": 0.09462399035692215, + "kept_token_count": 654.621875, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9918420970439911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157887309789658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.246533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.52951965332034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14310, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.20003580302000046, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.015397450421005487, + "grad_norm": 0.1256519854068756, + "kept_token_count": 654.753125, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923174917697907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007682507112622261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7658660888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7233825683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14320, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.19928528368473053, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.013448131643235684, + "grad_norm": 0.12688979506492615, + "kept_token_count": 655.240625, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9916713178157807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008328694384545088, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.55675048828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23839416503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14330, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20008975267410278, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.017861466389149427, + "grad_norm": 0.08452994376420975, + "kept_token_count": 654.2453125, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9924662351608277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075337671674788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3474822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.23781433105466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14340, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.20004441142082213, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.012672074465081095, + "grad_norm": 0.08418715000152588, + "kept_token_count": 654.540625, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928327262401581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007167277950793504, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5964324951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.77777099609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19908147007226945, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.01466857772320509, + "grad_norm": 0.08922840654850006, + "kept_token_count": 655.290625, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9921801924705506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007819793093949557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.6561798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.31340942382815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14360, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19943394809961318, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.013039224501699209, + "grad_norm": 0.08215352892875671, + "kept_token_count": 655.4, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9922551035881042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007744898553937674, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.7204986572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8608001708984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14370, + "valid_token_count": 818.6671875 + }, + { + "actual_token_mask_ratio": 0.2006600096821785, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.014750906638801098, + "grad_norm": 0.10468513518571854, + "kept_token_count": 654.321875, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922443270683289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007755671441555023, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.436181640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.51832580566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14380, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.200833697617054, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.015887431520968675, + "grad_norm": 0.09758216142654419, + "kept_token_count": 653.771875, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.02, + "mgd_cos_sim": 0.991817957162857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008182043675333261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.28480224609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1529174804688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14390, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19980664998292924, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.013860188703984023, + "grad_norm": 0.10388769954442978, + "kept_token_count": 654.725, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9925546228885651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007445370871573686, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.5794647216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5052734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14400, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.20033104866743087, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.0137617870233953, + "grad_norm": 0.11702626943588257, + "kept_token_count": 654.2453125, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9924333572387696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007566630188375711, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.532568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5312072753906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14410, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.20013203471899033, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.012512023700401187, + "grad_norm": 0.07190486788749695, + "kept_token_count": 654.3609375, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9922780156135559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007721981965005398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.42034301757815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.39686279296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14420, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20009908080101013, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.014399590808898211, + "grad_norm": 0.11773445457220078, + "kept_token_count": 654.33125, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9916744530200958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008325544651597738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.18027954101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8537353515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14430, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.2004934534430504, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.016225079260766506, + "grad_norm": 0.12098913639783859, + "kept_token_count": 654.1890625, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9926557302474975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00734425475820899, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.79168701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6008270263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14440, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2001885011792183, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.013040497712790967, + "grad_norm": 0.1325824111700058, + "kept_token_count": 654.5625, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9928923666477203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007107636053115129, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.30324401855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12864379882814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14450, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20003909021615982, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.013698899373412132, + "grad_norm": 0.11500377207994461, + "kept_token_count": 654.253125, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9927446484565735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00725534139201045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2297058105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86692810058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14460, + "valid_token_count": 817.8609375 + }, + { + "actual_token_mask_ratio": 0.19913756102323532, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.013971994258463382, + "grad_norm": 0.11751194298267365, + "kept_token_count": 655.5078125, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.992503547668457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007496444601565599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6698944091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8868133544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14470, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20016278326511383, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.012357882130891085, + "grad_norm": 0.11730290204286575, + "kept_token_count": 654.7234375, + "learning_rate": 1.709469158221008e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9926696598529816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007330338284373283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.98235778808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1637725830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14480, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.19941753000020981, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.01436674389988184, + "grad_norm": 0.07863584905862808, + "kept_token_count": 655.028125, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9917987704277038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008201240468770266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.77647705078124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.04689025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14490, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20052070915699005, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.013715655449777842, + "grad_norm": 0.08449644595384598, + "kept_token_count": 654.3546875, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9922694265842438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007730556838214398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.962646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1948486328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14500, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19905263483524321, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.016073678340762853, + "grad_norm": 0.10688906908035278, + "kept_token_count": 655.40625, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9922904551029206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007709543313831091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.68666687011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.45685119628905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14510, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19944703727960586, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.014274864643812179, + "grad_norm": 0.12476403266191483, + "kept_token_count": 655.2578125, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9923003792762757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007699615694582463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.0848846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8149078369141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14520, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.19910471588373185, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.014747971855103969, + "grad_norm": 0.10841400176286697, + "kept_token_count": 654.9734375, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9923227190971374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076772751286625866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.9911804199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.01044311523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14530, + "valid_token_count": 817.80625 + }, + { + "actual_token_mask_ratio": 0.20058561116456985, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.013753323908895254, + "grad_norm": 0.11029664427042007, + "kept_token_count": 654.1640625, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.992557805776596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007442193944007159, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1305450439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.20008544921876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14540, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.19958312511444093, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.015529712289571762, + "grad_norm": 0.10450642555952072, + "kept_token_count": 654.94375, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920989096164703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007901083771139383, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.0511138916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.97837524414064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14550, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19965713918209077, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.015194022748619317, + "grad_norm": 0.09703677892684937, + "kept_token_count": 654.78125, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9924802303314209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007519753836095333, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.6477569580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6524963378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14560, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20010821521282196, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.01402659947052598, + "grad_norm": 0.0839858129620552, + "kept_token_count": 654.5125, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9923097431659699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076902519911527635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.03191833496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65943298339846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14570, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19987961947917937, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.014125031651929021, + "grad_norm": 0.11110808700323105, + "kept_token_count": 654.6859375, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9919073522090912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00809264723211527, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3345947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.64324340820315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14580, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20045447647571563, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.014184455666691066, + "grad_norm": 0.14963696897029877, + "kept_token_count": 654.2265625, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9923014342784882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00769857233390212, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.4902404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9727020263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14590, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20098277181386948, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.013179113762453199, + "grad_norm": 0.08208255469799042, + "kept_token_count": 653.784375, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930271506309509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006972866877913475, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.3024627685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3722808837891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14600, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.19863636344671248, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.014665715396404266, + "grad_norm": 0.09883919358253479, + "kept_token_count": 655.515625, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9923634648323059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007636553887277841, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.02231750488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9986541748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14610, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.19991346448659897, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.013650319632142782, + "grad_norm": 0.1243474930524826, + "kept_token_count": 654.5421875, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992196899652481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007803100161254406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3035430908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5389343261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14620, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.19989343285560607, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.013555476535111665, + "grad_norm": 0.08963260799646378, + "kept_token_count": 654.746875, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9925258934497834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007474107015877962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.416455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.91285400390626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.20105236917734146, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.014260416105389594, + "grad_norm": 0.11488079279661179, + "kept_token_count": 653.703125, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9926723122596741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007327690999954939, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6214965820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.07464294433595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14640, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19974734932184218, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.013571920804679394, + "grad_norm": 0.13231340050697327, + "kept_token_count": 654.9015625, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9916050791740417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008394922781735658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.5339660644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8006927490234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.19967075437307358, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.013936142809689045, + "grad_norm": 0.13082122802734375, + "kept_token_count": 654.93125, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9921808779239655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007819124311208726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1208129882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0118865966797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14660, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19911685585975647, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.017412167321890593, + "grad_norm": 0.10183922201395035, + "kept_token_count": 655.4453125, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9917857110500335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008214293606579303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.9393646240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.79649963378904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14670, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.1993188440799713, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.015283996984362603, + "grad_norm": 0.11084357649087906, + "kept_token_count": 655.4046875, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9920753240585327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007924672402441501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7671813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.88328552246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14680, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.19946122616529466, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.013510787487030029, + "grad_norm": 0.10526856780052185, + "kept_token_count": 655.0078125, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.99262033700943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007379651069641113, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1471771240234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14690, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19950858056545256, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.016390954796224832, + "grad_norm": 0.11249802261590958, + "kept_token_count": 655.1390625, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9924935221672058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007506464514881373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2779144287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.5529022216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14700, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19983019232749938, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.013170109502971173, + "grad_norm": 0.09538105875253677, + "kept_token_count": 654.9359375, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.992162823677063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007837188430130482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.98846130371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0458038330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14710, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.20035169422626495, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.014936166908591985, + "grad_norm": 0.11822686344385147, + "kept_token_count": 654.7015625, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9919441282749176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008055882714688777, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.2550323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7836059570312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14720, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.20053808838129045, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.016886528953909873, + "grad_norm": 0.13350611925125122, + "kept_token_count": 654.2609375, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9923143267631531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007685663085430861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.6353332519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1136047363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14730, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.20026795715093612, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.013523311959579587, + "grad_norm": 0.10423565655946732, + "kept_token_count": 654.1796875, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926809370517731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007319056242704391, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6678405761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6151580810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14740, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.20036479383707045, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.0131589001044631, + "grad_norm": 0.10950159281492233, + "kept_token_count": 654.146875, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.992624431848526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007375574484467506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.3776885986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.17486572265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14750, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.20022102296352387, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.015443526301532983, + "grad_norm": 0.11583317816257477, + "kept_token_count": 654.7984375, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.992497444152832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007502551469951868, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 455.1651214599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2840545654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14760, + "valid_token_count": 818.7328125 + }, + { + "actual_token_mask_ratio": 0.20051494240760803, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.013984751608222722, + "grad_norm": 0.09834127128124237, + "kept_token_count": 654.5390625, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927618622779846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007238122541457415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.72313537597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7782287597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14770, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.20038034319877623, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.01434384835883975, + "grad_norm": 0.10466045886278152, + "kept_token_count": 654.25, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922568440437317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007743169739842415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7823181152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.61944274902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14780, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2004537835717201, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.01380269331857562, + "grad_norm": 0.13862065970897675, + "kept_token_count": 654.2375, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9918517827987671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008148225396871567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4814392089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2906555175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14790, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20062438100576402, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.015130044613033532, + "grad_norm": 0.10150065273046494, + "kept_token_count": 653.9921875, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9927392363548279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007260763086378574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0801300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.35333557128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14800, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20033951997756957, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.015049932524561882, + "grad_norm": 0.0962323322892189, + "kept_token_count": 654.6015625, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922438859939575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007756118196994066, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.94468994140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.86072692871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14810, + "valid_token_count": 818.5984375 + }, + { + "actual_token_mask_ratio": 0.20067577362060546, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.012261581234633922, + "grad_norm": 0.08063244074583054, + "kept_token_count": 654.2421875, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9927315473556518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007268444821238518, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1404998779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.60466918945315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14820, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20059247612953185, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.01606064196676016, + "grad_norm": 0.09097486734390259, + "kept_token_count": 654.25, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9916110217571259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008388980478048324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3078338623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7386077880859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14830, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.19860336184501648, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.015018610283732414, + "grad_norm": 0.1253758668899536, + "kept_token_count": 655.5796875, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924519479274749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075480474159121515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.78589782714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5647674560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14840, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20036105066537857, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.012251659110188485, + "grad_norm": 0.07771921902894974, + "kept_token_count": 654.3125, + "learning_rate": 1.648732828427283e-05, + "loss": 0.016, + "mgd_cos_sim": 0.992439478635788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075605249963700775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6439178466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7490509033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14850, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19997481107711793, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.015043113753199578, + "grad_norm": 0.10494618862867355, + "kept_token_count": 654.74375, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.992459911108017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007540093082934618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.308984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94317932128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14860, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20063092857599257, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.014992510434240103, + "grad_norm": 0.10176098346710205, + "kept_token_count": 654.1359375, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9919806957244873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008019303809851408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1099517822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4304718017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14870, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.20061393827199936, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.015901964157819748, + "grad_norm": 0.12999476492404938, + "kept_token_count": 654.071875, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9921050608158112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007894933503121138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.5228668212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2002410888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14880, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20028865039348603, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.014353981707245111, + "grad_norm": 0.1040656566619873, + "kept_token_count": 654.565625, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924254775047302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00757452379912138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2305480957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7318878173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14890, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.1997307702898979, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.01722253244370222, + "grad_norm": 0.132376566529274, + "kept_token_count": 655.028125, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923097670078278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007690238580107689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.60040588378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.65980529785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14900, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20056479722261428, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.015576052479445934, + "grad_norm": 0.10829359292984009, + "kept_token_count": 654.0296875, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9923018038272857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007698186673223972, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.39052429199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8339782714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14910, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.19990361630916595, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.013721797335892915, + "grad_norm": 0.0949428454041481, + "kept_token_count": 654.684375, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926557123661042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007344288285821676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.32790222167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2994323730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14920, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19990011304616928, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.01694824593141675, + "grad_norm": 0.15392857789993286, + "kept_token_count": 654.7015625, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9921460807323456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007853916473686695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.364697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5452453613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14930, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.19975428730249406, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.015596954245120287, + "grad_norm": 0.10765141248703003, + "kept_token_count": 654.9125, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9923038959503174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007696104701608419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.02965087890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.95013122558595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14940, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.19916850328445435, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.01401574732735753, + "grad_norm": 0.13621969521045685, + "kept_token_count": 655.3140625, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9922573864459991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007742614299058914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.9202423095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.45888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14950, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.20044130980968475, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.012664173962548375, + "grad_norm": 0.09753995388746262, + "kept_token_count": 654.465625, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9924662709236145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007533724140375853, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4495910644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.95314331054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14960, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19931121170520782, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.014755783043801785, + "grad_norm": 0.09345945715904236, + "kept_token_count": 655.5765625, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922867178916931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00771328154951334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.59193420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8746307373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14970, + "valid_token_count": 818.7609375 + }, + { + "actual_token_mask_ratio": 0.19934353232383728, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.014761699363589287, + "grad_norm": 0.0888097733259201, + "kept_token_count": 654.9515625, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9928048729896546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007195129990577698, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.74451599121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07082214355466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14980, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.1986263796687126, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.013904635794460773, + "grad_norm": 0.1186656728386879, + "kept_token_count": 655.6953125, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927407205104828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072592804208397865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1479827880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1197265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 14990, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20063198506832122, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.014658290706574918, + "grad_norm": 0.13433144986629486, + "kept_token_count": 654.190625, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926442503929138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007355764973908663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3253875732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3630828857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15000, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20054722875356673, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.01629817150533199, + "grad_norm": 0.115815170109272, + "kept_token_count": 654.35, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9925861120223999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007413895428180694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9158935546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.31610107421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15010, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19995155781507493, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.015545407030731439, + "grad_norm": 0.12496446818113327, + "kept_token_count": 654.7515625, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927378833293915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007262125704437494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.0373870849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2178436279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15020, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2001399427652359, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.013070321083068848, + "grad_norm": 0.12466227263212204, + "kept_token_count": 654.54375, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9925949513912201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405044510960579, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1528045654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7202606201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15030, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.19967535138130188, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.013111583516001701, + "grad_norm": 0.09993920475244522, + "kept_token_count": 654.95, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992517226934433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007482769805938006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0837371826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8987579345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19969488084316253, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.013830081652849912, + "grad_norm": 0.11214316636323929, + "kept_token_count": 655.0078125, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9930467307567596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953274458646774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5359313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68885498046876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15050, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.19945504367351533, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.014191313553601504, + "grad_norm": 0.08089388906955719, + "kept_token_count": 655.2015625, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9933390617370605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006660935189574957, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5985504150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7426513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15060, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.19929929971694946, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.015566583536565304, + "grad_norm": 0.10703578591346741, + "kept_token_count": 655.0234375, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927428722381592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007257136981934309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8411804199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8706512451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15070, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.19941385835409164, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.01715646917000413, + "grad_norm": 0.12551911175251007, + "kept_token_count": 655.1875, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9924764692783355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007523529790341854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5214080810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9736297607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15080, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.20022946298122407, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.016596696339547634, + "grad_norm": 0.12416010349988937, + "kept_token_count": 654.3125, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992416375875473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007583630830049514, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.3276611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.53834533691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15090, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19932845681905748, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.014946040231734514, + "grad_norm": 0.10784045606851578, + "kept_token_count": 655.053125, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.992417985200882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007582017220556736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.8533538818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0538055419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15100, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19976873099803924, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.01596239530481398, + "grad_norm": 0.0925007164478302, + "kept_token_count": 654.5984375, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923932552337646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007606739923357963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5511505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.62465209960936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15110, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19940199404954911, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.014577221032232046, + "grad_norm": 0.13954545557498932, + "kept_token_count": 655.2875, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9926802396774292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007319765724241733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 467.2178192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5866455078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15120, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20009392946958543, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.012556830886751413, + "grad_norm": 0.1005915030837059, + "kept_token_count": 654.7671875, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9921561479568481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843856420367955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.62860412597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62726135253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15130, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.20007209330797196, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.016068992763757707, + "grad_norm": 0.16142334043979645, + "kept_token_count": 654.7359375, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9917891204357148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210873883217572, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.15562744140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3841186523438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15140, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.19909486770629883, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.013563570566475391, + "grad_norm": 0.08846838772296906, + "kept_token_count": 655.146875, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9924744129180908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007525577954947948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6305725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4103149414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15150, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.19991685450077057, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.01414280771277845, + "grad_norm": 0.13334417343139648, + "kept_token_count": 654.4390625, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9916371166706085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008362895902246236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.30140380859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22503967285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15160, + "valid_token_count": 817.965625 + }, + { + "actual_token_mask_ratio": 0.20006588250398635, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.015231456886976958, + "grad_norm": 0.09518994390964508, + "kept_token_count": 654.4203125, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9921145796775818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007885421719402074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1822937011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.20146484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15170, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.19905905574560165, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.015040924400091171, + "grad_norm": 0.11087750643491745, + "kept_token_count": 655.5546875, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9921678483486176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007832148391753436, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.16639099121096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.053466796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15180, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.19961659014225006, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.01689693722873926, + "grad_norm": 0.10929188132286072, + "kept_token_count": 654.903125, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9922321498394012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007767841964960098, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.97765197753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9041748046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15190, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1997348740696907, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.01476886086165905, + "grad_norm": 0.1001935750246048, + "kept_token_count": 654.6640625, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9926098465919495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007390158530324697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.3091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2979034423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15200, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.19905715435743332, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.013643487635999919, + "grad_norm": 0.10131530463695526, + "kept_token_count": 655.5390625, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9927502453327179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072497445158660415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.36871948242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1454711914063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15210, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.1998026356101036, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.014320275373756885, + "grad_norm": 0.09441713243722916, + "kept_token_count": 654.5578125, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922686994075776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007731314375996589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.6217346191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03026123046874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15220, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.20021132230758668, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.015426542051136494, + "grad_norm": 0.11286456882953644, + "kept_token_count": 654.4484375, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924906015396118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075094115920364855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0391510009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2028106689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15230, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.20007635205984114, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.015372548997402192, + "grad_norm": 0.13557425141334534, + "kept_token_count": 654.6484375, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9920889377593994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007911065500229597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5755126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.64955444335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.2000168189406395, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.014618225581943988, + "grad_norm": 0.11917629092931747, + "kept_token_count": 654.4015625, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9925003588199616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007499650865793228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 466.2092620849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22490234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15250, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19930774420499803, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.01390569182112813, + "grad_norm": 0.10522181540727615, + "kept_token_count": 655.090625, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9924000144004822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007599994726479054, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.89134826660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.01792602539064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15260, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.20032358169555664, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.017837081197649242, + "grad_norm": 0.09830066561698914, + "kept_token_count": 654.290625, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9922728419303894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007727140933275223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0561065673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8752838134766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15270, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19996761083602904, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.015046033542603254, + "grad_norm": 0.08773671090602875, + "kept_token_count": 654.734375, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927905917167663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209415640681982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7487457275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6990509033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15280, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20048808604478835, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.014066244103014469, + "grad_norm": 0.10174605995416641, + "kept_token_count": 654.1734375, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927000224590301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299988996237516, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6715545654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2462188720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15290, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.1989235132932663, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.01554102418012917, + "grad_norm": 0.1256735622882843, + "kept_token_count": 655.50625, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920315384864807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007968462817370892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2476104736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4017364501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15300, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.20033001750707627, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.013296126993373037, + "grad_norm": 0.09801064431667328, + "kept_token_count": 654.3671875, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9916731417179108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008326857164502145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.70860290527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0565612792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15310, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2000100702047348, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.011914188973605633, + "grad_norm": 0.088255375623703, + "kept_token_count": 654.68125, + "learning_rate": 1.571237742782112e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9919027209281921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008097257278859616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.04454345703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1487579345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15320, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19977468252182007, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.014376156777143479, + "grad_norm": 0.09498811513185501, + "kept_token_count": 654.5640625, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9925506830215454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007449305243790149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 471.1850280761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.43695068359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15330, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.20001538693904877, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.01276101404801011, + "grad_norm": 0.11139792948961258, + "kept_token_count": 654.590625, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928972005844117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007102784607559443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.2878051757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6396514892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15340, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19963209331035614, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.014208250772207975, + "grad_norm": 0.10520375519990921, + "kept_token_count": 654.8296875, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927789688110351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007221040688455105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.74619750976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23024597167966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15350, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.1994536116719246, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.01512953294441104, + "grad_norm": 0.1105237826704979, + "kept_token_count": 655.2953125, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9927285671234131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007271421793848276, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.0203521728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.79603881835936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15360, + "valid_token_count": 818.5578125 + }, + { + "actual_token_mask_ratio": 0.20032237768173217, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.01597375376150012, + "grad_norm": 0.11094889044761658, + "kept_token_count": 654.440625, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923030436038971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007696945406496525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.10927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.08868713378905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15370, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.1988495647907257, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.014031869452446699, + "grad_norm": 0.10297854244709015, + "kept_token_count": 655.4625, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9922060132026672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007793995179235935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.4516662597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0607147216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15380, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.1996268853545189, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.015964743122458457, + "grad_norm": 0.12416701018810272, + "kept_token_count": 654.9046875, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923170864582062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007682915311306715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.13356018066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1678833007812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15390, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20076964646577836, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.014229773543775082, + "grad_norm": 0.10433350503444672, + "kept_token_count": 654.23125, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9924202978610992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007579702138900757, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7327545166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.72149658203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15400, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.1998808741569519, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.01267611226066947, + "grad_norm": 0.09788978099822998, + "kept_token_count": 654.678125, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.992359870672226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00764013035222888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.78880310058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2628173828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15410, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.199920129776001, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.016408472787588835, + "grad_norm": 0.10541977733373642, + "kept_token_count": 654.790625, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9925295591354371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007470445521175861, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5168884277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.23841552734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15420, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20026157349348067, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.013937100302428007, + "grad_norm": 0.10988323390483856, + "kept_token_count": 654.4796875, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992172920703888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007827069796621799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8814727783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5736022949219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15430, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.20006785094738005, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.016352040879428386, + "grad_norm": 0.11829689890146255, + "kept_token_count": 654.7015625, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9921800851821899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007819922640919685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9076232910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98460998535154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15440, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.20011701732873916, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.013391584903001786, + "grad_norm": 0.08918960392475128, + "kept_token_count": 654.628125, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9920861899852753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007913809176534415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2110656738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15450, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19913629740476607, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.013911249488592148, + "grad_norm": 0.09060710668563843, + "kept_token_count": 655.2, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9924507975578308, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007549205236136914, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3116851806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2502471923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15460, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.19983392655849458, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.011996412416920066, + "grad_norm": 0.09694827347993851, + "kept_token_count": 654.815625, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9925201594829559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007479824963957072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.79371948242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8778900146484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.20062171518802643, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.01590678934007883, + "grad_norm": 0.09530946612358093, + "kept_token_count": 653.659375, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9922950804233551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007704928424209357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2090728759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9556182861328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15480, + "valid_token_count": 817.709375 + }, + { + "actual_token_mask_ratio": 0.19958228915929793, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.014836054388433694, + "grad_norm": 0.08671139180660248, + "kept_token_count": 654.846875, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9924250066280365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075749937444925305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5017486572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.83380432128905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15490, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19967132806777954, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.015900356601923703, + "grad_norm": 0.1051744818687439, + "kept_token_count": 654.76875, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9921585023403168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007841501105576753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9705047607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0604278564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15500, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20035423040390016, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.013591342605650425, + "grad_norm": 0.09870477765798569, + "kept_token_count": 654.5140625, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9921791017055511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007820906676352024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.8906188964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3451293945312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15510, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.19955047816038132, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.015395336132496596, + "grad_norm": 0.1322975605726242, + "kept_token_count": 654.84375, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924934387207032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075065714307129385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.70728454589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2750213623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15520, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.1995609074831009, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.012903024302795529, + "grad_norm": 0.14731906354427338, + "kept_token_count": 654.865625, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9930651903152465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006934796925634146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.15382995605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1364715576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15530, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.20054754316806794, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.014357380475848914, + "grad_norm": 0.11435436457395554, + "kept_token_count": 653.81875, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9919573545455933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00804264871403575, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.26336364746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.13662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15540, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.19891889542341232, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.01546187661588192, + "grad_norm": 0.08628476411104202, + "kept_token_count": 655.771875, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924737274646759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007526275515556335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0946441650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1973480224609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15550, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.19993031322956084, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.015573786478489638, + "grad_norm": 0.07718268036842346, + "kept_token_count": 654.5203125, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9923133730888367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007686623279005289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.77728881835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.64948120117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15560, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.19930684715509414, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.01680163452401757, + "grad_norm": 0.12629951536655426, + "kept_token_count": 655.2109375, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9921747982501984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007825197000056506, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0828399658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5206726074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15570, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20060293674468993, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.015255587454885245, + "grad_norm": 0.08442624658346176, + "kept_token_count": 654.5640625, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9927135705947876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007286409568041563, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5532928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9528411865234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15580, + "valid_token_count": 818.815625 + }, + { + "actual_token_mask_ratio": 0.2008829414844513, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.013586768321692944, + "grad_norm": 0.10908138751983643, + "kept_token_count": 653.9484375, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9921567022800446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843302749097347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.29334716796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.11835327148435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15590, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2002217024564743, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.015653010457754135, + "grad_norm": 0.10741658508777618, + "kept_token_count": 654.725, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9918534159660339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008146590646356344, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.29583129882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1236328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15600, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20007692128419877, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.01439470062032342, + "grad_norm": 0.11724165827035904, + "kept_token_count": 654.8640625, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9924297094345093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007570300716906786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.66387634277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4301818847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15610, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.19936659932136536, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.013048662897199392, + "grad_norm": 0.11434279382228851, + "kept_token_count": 655.1609375, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9922340989112854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007765890657901764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.393994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.18332824707034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15620, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2006736844778061, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.012971073016524315, + "grad_norm": 0.10397969186306, + "kept_token_count": 654.0984375, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9925623714923859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007437611557543277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2834442138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.32762756347654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15630, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.20082086473703384, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.01487322561442852, + "grad_norm": 0.10404318571090698, + "kept_token_count": 654.134375, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9920635938644409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00793642457574606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6590850830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.667431640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15640, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.1995278164744377, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.01336024459451437, + "grad_norm": 0.0976315438747406, + "kept_token_count": 654.9953125, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9925757765769958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00742422379553318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.9957305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5386688232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15650, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.1995289668440819, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.015121184382587671, + "grad_norm": 0.10425952076911926, + "kept_token_count": 655.2109375, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922602236270904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007739776093512773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.86207275390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9568267822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15660, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.20086174309253693, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.01283876020461321, + "grad_norm": 0.10056081414222717, + "kept_token_count": 654.11875, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9924235284328461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007576477900147438, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.7896301269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.100634765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15670, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.19998023360967637, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.01333597325719893, + "grad_norm": 0.11547116935253143, + "kept_token_count": 654.5953125, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9922854840755463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007714512944221497, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.7481384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1326843261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2005072385072708, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.012855565547943116, + "grad_norm": 0.11748925596475601, + "kept_token_count": 654.16875, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9925115287303925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00748846735805273, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.31324768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68499755859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15690, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.20044149756431578, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.013740980811417104, + "grad_norm": 0.10309725999832153, + "kept_token_count": 654.11875, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9918330430984497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008166961837559938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.99046020507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7535461425781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15700, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.20058625042438508, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.014428711635991931, + "grad_norm": 0.09549281001091003, + "kept_token_count": 653.9015625, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9923282325267792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00767176290974021, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.38889465332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.74168395996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15710, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.20007830262184143, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013212104793637991, + "grad_norm": 0.12249314785003662, + "kept_token_count": 654.6109375, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926359236240387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073640717193484305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.97092590332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.53782043457034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19940713793039322, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.014198062662035226, + "grad_norm": 0.10513834655284882, + "kept_token_count": 655.159375, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9923834085464478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00761659424751997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9949920654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5162078857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15730, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19975989311933517, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.013402853859588504, + "grad_norm": 0.12804999947547913, + "kept_token_count": 654.6734375, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9923539578914642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076460462994873525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.274755859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6988067626953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15740, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20045454055070877, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.015003645233809948, + "grad_norm": 0.08866464346647263, + "kept_token_count": 653.98125, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9929759621620178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007024036440998316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5110565185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.640478515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15750, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.19976688772439957, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.014398236572742463, + "grad_norm": 0.14136838912963867, + "kept_token_count": 654.75625, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.991940951347351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008059044554829597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7069915771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.78208312988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15760, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2003748819231987, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.014798145135864615, + "grad_norm": 0.10595807433128357, + "kept_token_count": 654.2453125, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9929096043109894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00709039643406868, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.54380187988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7030792236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20054296851158143, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.014798388117924333, + "grad_norm": 0.13338272273540497, + "kept_token_count": 653.9265625, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9923759460449219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007624049577862025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.05101318359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.44935607910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15780, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.19928345680236817, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.0150700394064188, + "grad_norm": 0.10547132045030594, + "kept_token_count": 655.18125, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926869094371795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00731308963149786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.7661071777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.70884704589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2003428429365158, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.015377934463322163, + "grad_norm": 0.12174071371555328, + "kept_token_count": 654.56875, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924048244953155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007595181558281183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.23479614257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0839813232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15800, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.20102299600839615, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.014142418280243874, + "grad_norm": 0.10064984858036041, + "kept_token_count": 653.6234375, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9922803401947021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077196724712848665, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.48460998535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68639221191404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15810, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19993662983179092, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.01437312737107277, + "grad_norm": 0.1369781196117401, + "kept_token_count": 654.5546875, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9924235880374909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007576418295502663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.269970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18650207519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15820, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.20007880181074142, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.013148951716721058, + "grad_norm": 0.11588425934314728, + "kept_token_count": 654.56875, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930375039577484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00696249483153224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5345825195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8301513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15830, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19949963092803955, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.013996627274900674, + "grad_norm": 0.14168238639831543, + "kept_token_count": 655.040625, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9921163260936737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007883676327764988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.84531860351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.964697265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15840, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.20046541839838028, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.015381793584674596, + "grad_norm": 0.09924844652414322, + "kept_token_count": 654.1796875, + "learning_rate": 1.483630973711214e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9927135109901428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072865084744989875, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3263427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7159851074219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15850, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.20036772191524505, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.01774407932534814, + "grad_norm": 0.11044631898403168, + "kept_token_count": 654.3671875, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9922813653945923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007718628086149692, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.60979614257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.79217834472655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15860, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2009532332420349, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.014788083732128143, + "grad_norm": 0.09708642214536667, + "kept_token_count": 653.75625, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9926134288311005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073865680024027824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0869506835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7984283447266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15870, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19949121475219728, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.014737191330641507, + "grad_norm": 0.10222087055444717, + "kept_token_count": 655.046875, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9924855589866638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00751444036141038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.71409912109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0079010009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15880, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19964174181222916, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.016609782353043556, + "grad_norm": 0.12225772440433502, + "kept_token_count": 654.6125, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9926303744316101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073696085251867775, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.55031127929686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4179382324219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15890, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.20006423741579055, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.013505981676280499, + "grad_norm": 0.11848686635494232, + "kept_token_count": 654.440625, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.992504495382309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007495500333607197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.98520812988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.53065185546876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15900, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.20051637887954712, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.013899741042405368, + "grad_norm": 0.13722015917301178, + "kept_token_count": 654.4453125, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9922885477542878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077114621177315715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.07521362304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7405303955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15910, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.1996808722615242, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.013924928568303584, + "grad_norm": 0.12721829116344452, + "kept_token_count": 655.109375, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9922636330127717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077363581396639345, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.20859680175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03490295410154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15920, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.20002057999372483, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.014758492074906825, + "grad_norm": 0.12379032373428345, + "kept_token_count": 654.7875, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9916330456733704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008366944827139377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.6599548339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.33863220214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15930, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.20070318430662154, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.013302583293989301, + "grad_norm": 0.11397906392812729, + "kept_token_count": 654.225, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930602610111237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939734425395727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6634460449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6169464111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15940, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.19965252578258513, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.01594677669927478, + "grad_norm": 0.12973132729530334, + "kept_token_count": 655.0296875, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9928402543067932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007159745786339045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.27586669921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3088836669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15950, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2000096023082733, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.013322607427835465, + "grad_norm": 0.08697729557752609, + "kept_token_count": 654.7796875, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9924201428890228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007579874154180289, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.533251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8519317626953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15960, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.19976383149623872, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.01423139963299036, + "grad_norm": 0.1097777932882309, + "kept_token_count": 655.021875, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9929037630558014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007096241414546967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1746795654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8057830810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15970, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.20073627680540085, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.015434309840202332, + "grad_norm": 0.1489787995815277, + "kept_token_count": 654.0078125, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925211548805237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007478856574743986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8536895751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5264373779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15980, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.200627663731575, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.014475621934980153, + "grad_norm": 0.12193384766578674, + "kept_token_count": 654.2203125, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9925308585166931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007469144184142351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1438690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.287109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 15990, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.2007328003644943, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.013107182271778584, + "grad_norm": 0.09527503699064255, + "kept_token_count": 654.2640625, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926559388637543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007344065140932799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.7998504638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33787231445314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16000, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.19974658787250518, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.012568134069442748, + "grad_norm": 0.11197640746831894, + "kept_token_count": 654.6375, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.992227029800415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007772970478981733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4320434570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6056823730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16010, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.2003291517496109, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.015275237895548344, + "grad_norm": 0.13858065009117126, + "kept_token_count": 654.1859375, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9921510398387909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007848972082138061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0115051269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0037536621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16020, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.1993207722902298, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.012474529212340713, + "grad_norm": 0.09997843950986862, + "kept_token_count": 655.3703125, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9924196839332581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007580308429896832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.31910095214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.30068054199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16030, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.20033077746629716, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.013756969105452299, + "grad_norm": 0.08226065337657928, + "kept_token_count": 654.4015625, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9925401210784912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007459875196218491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.70252990722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.683984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16040, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19910001158714294, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.01478542210534215, + "grad_norm": 0.11383795738220215, + "kept_token_count": 655.2375, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9924808979034424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007519115321338176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1924163818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5812255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16050, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.1998681142926216, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.015671302983537316, + "grad_norm": 0.17336872220039368, + "kept_token_count": 654.475, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9926823914051056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007317616324871779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.16441345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3614044189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16060, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.20037236213684081, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.01370697021484375, + "grad_norm": 0.11589031666517258, + "kept_token_count": 654.440625, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9923872649669647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076127382926642895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.69736633300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3520843505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2004951775074005, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.014956585224717855, + "grad_norm": 0.09608309715986252, + "kept_token_count": 654.034375, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9925539314746856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007446067873388529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8105194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4697601318359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16080, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.19979609847068786, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.017522635404020548, + "grad_norm": 0.09360668808221817, + "kept_token_count": 654.5375, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.992650818824768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007349181268364191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.76189575195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.30213317871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16090, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.200067038834095, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.0115456182975322, + "grad_norm": 0.10530200600624084, + "kept_token_count": 654.5453125, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9921003162860871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007899685576558114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.06125793457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30592956542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16100, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19989480078220367, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.013551314454525709, + "grad_norm": 0.09562937170267105, + "kept_token_count": 654.534375, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9924639403820038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007536064833402634, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.851513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8890899658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16110, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.20096912384033203, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.013518446870148183, + "grad_norm": 0.09691964089870453, + "kept_token_count": 653.859375, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9922323763370514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007767617609351874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2261993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7683074951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16120, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2005681276321411, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.013660658895969392, + "grad_norm": 0.1259370595216751, + "kept_token_count": 654.071875, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9923382997512817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007661694753915071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5314697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.05430297851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16130, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.1989060401916504, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.014563164673745632, + "grad_norm": 0.09668731689453125, + "kept_token_count": 655.45, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9924173891544342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007582601625472307, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.5675445556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.16851501464845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16140, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.19996558725833893, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.013904789928346872, + "grad_norm": 0.1143215224146843, + "kept_token_count": 654.2359375, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927269756793976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007273032795637846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3521728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.62122497558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16150, + "valid_token_count": 817.7609375 + }, + { + "actual_token_mask_ratio": 0.20099978893995285, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.012842956744134427, + "grad_norm": 0.10670895129442215, + "kept_token_count": 653.5390625, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928046584129333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071953476406633856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1147155761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0167663574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16160, + "valid_token_count": 817.946875 + }, + { + "actual_token_mask_ratio": 0.20002889931201934, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.013738846965134143, + "grad_norm": 0.10968992114067078, + "kept_token_count": 654.7390625, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9924525856971741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007547412440180779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 462.3228424072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.47734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.20089684575796127, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.015026859845966101, + "grad_norm": 0.14723263680934906, + "kept_token_count": 653.9359375, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922569215297699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007743061427026987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 457.2585479736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.15894775390626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16180, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.2006015881896019, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.014741781912744046, + "grad_norm": 0.15246349573135376, + "kept_token_count": 654.128125, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9931059181690216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006894082576036453, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5035400390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.13608703613284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16190, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.19980139881372452, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.016442238166928292, + "grad_norm": 0.0947137102484703, + "kept_token_count": 654.6546875, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9921422302722931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007857787795364857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.54307250976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.13213806152345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16200, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20010113418102266, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.015255713369697332, + "grad_norm": 0.09760918468236923, + "kept_token_count": 654.534375, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922549247741699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007745076809078455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 464.13645629882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.40452880859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16210, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1991448238492012, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.015140809677541256, + "grad_norm": 0.10601765662431717, + "kept_token_count": 655.25, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927991569042206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007200844120234251, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.59007873535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.88409423828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16220, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.1999746099114418, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.01238424303010106, + "grad_norm": 0.08436340093612671, + "kept_token_count": 654.584375, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928370416164398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00716296648606658, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.49830017089846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.682470703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.20010669827461242, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.011209515389055014, + "grad_norm": 0.10322567075490952, + "kept_token_count": 654.75, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9928893804550171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007110633328557015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.4202575683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3505859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16240, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.2007969707250595, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.015323753654956817, + "grad_norm": 0.11475387215614319, + "kept_token_count": 654.0796875, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9930584669113159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941534951329232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.5945343017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.61016235351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16250, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.20052181482315062, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.014417791226878762, + "grad_norm": 0.10426861047744751, + "kept_token_count": 654.046875, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.992428320646286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007571681216359138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6854644775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2052276611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16260, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.19989830404520034, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.016360661946237087, + "grad_norm": 0.1375378966331482, + "kept_token_count": 654.878125, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9926871836185456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007312824763357639, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8664154052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5170837402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16270, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.20086168348789216, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.014422721695154905, + "grad_norm": 0.09864597767591476, + "kept_token_count": 654.0328125, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.992134028673172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007865967135876418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5797607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.79852905273435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16280, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20026351660490035, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.018651426117867232, + "grad_norm": 0.10387895256280899, + "kept_token_count": 654.3625, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9922796070575715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007720392197370529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.86805114746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5643280029297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19958838522434236, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.01540066348388791, + "grad_norm": 0.0966109186410904, + "kept_token_count": 655.2390625, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924444615840912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007555528916418553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1344970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10537109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16300, + "valid_token_count": 818.6171875 + }, + { + "actual_token_mask_ratio": 0.19881657510995865, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.013554634060710669, + "grad_norm": 0.12137100845575333, + "kept_token_count": 655.4015625, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926763355731965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007323652226477862, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.49566040039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84645690917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16310, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.19969532340765, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.014443787652999163, + "grad_norm": 0.11395060271024704, + "kept_token_count": 654.9984375, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9925596177577972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007440396491438151, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4237030029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.62803039550784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16320, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.20054271817207336, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.01297772079706192, + "grad_norm": 0.09094145894050598, + "kept_token_count": 654.23125, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9925955593585968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00740444753319025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.60523681640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0747406005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16330, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19956004917621611, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.01553448187187314, + "grad_norm": 0.0961504578590393, + "kept_token_count": 654.8625, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927562534809112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072437523864209655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.47132568359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8371948242187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16340, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.20001906901597977, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.013370249047875404, + "grad_norm": 0.12486676871776581, + "kept_token_count": 654.5703125, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9926239073276519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00737608689814806, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5271697998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.370458984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16350, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1998392477631569, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.015240762662142516, + "grad_norm": 0.15204836428165436, + "kept_token_count": 654.6625, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9925682723522187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074317330494523045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.3139221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.58311157226564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16360, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20030725002288818, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.012702596560120582, + "grad_norm": 0.0880771204829216, + "kept_token_count": 654.359375, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9929255783557892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007074423972517252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.47138061523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3534362792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16370, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20005386024713517, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.013667181413620711, + "grad_norm": 0.11326814442873001, + "kept_token_count": 654.51875, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926060914993287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007393907755613327, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1152404785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6949920654297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16380, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20064307302236556, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.01591798206791282, + "grad_norm": 0.11192581057548523, + "kept_token_count": 654.0515625, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9919339001178742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008066097926348447, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3088073730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03673095703124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16390, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19903123676776885, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.012889043148607015, + "grad_norm": 0.10533103346824646, + "kept_token_count": 655.3109375, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9928773939609528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071225928142666815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0480194091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7318969726563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16400, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20001566112041474, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.013852701894938946, + "grad_norm": 0.09798561036586761, + "kept_token_count": 654.6359375, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9929356813430786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007064318843185901, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3696990966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.65960998535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16410, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.19923897087574005, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.013498165179044008, + "grad_norm": 0.08255337178707123, + "kept_token_count": 655.2515625, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9924957394599915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007504267618060112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.7596954345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6602813720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.1991218999028206, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.014304418209940195, + "grad_norm": 0.08971309661865234, + "kept_token_count": 655.5140625, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925940275192261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405985705554485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.32135009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9352783203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16430, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2003044828772545, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.012509842030704021, + "grad_norm": 0.08123976737260818, + "kept_token_count": 654.48125, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9932157516479492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00678423885256052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.70518188476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2558258056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16440, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.20011185705661774, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.01500424137338996, + "grad_norm": 0.09849411994218826, + "kept_token_count": 654.465625, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922130167484283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077869811095297335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1833953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2580871582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16450, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19948963224887847, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.011721155606210232, + "grad_norm": 0.0911589190363884, + "kept_token_count": 655.25625, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9927575469017029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007242451142519713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.37290954589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5285705566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16460, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.20031227618455888, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.012847271654754877, + "grad_norm": 0.08799384534358978, + "kept_token_count": 654.5671875, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9925710618495941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007428958918899298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.7211639404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.0092712402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16470, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.20016409158706666, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.016562538035213948, + "grad_norm": 0.12601983547210693, + "kept_token_count": 654.4546875, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992366361618042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007633649371564388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.98163452148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1465728759766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16480, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.20060763657093048, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.014854627009481192, + "grad_norm": 0.13382847607135773, + "kept_token_count": 654.23125, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927412927150726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007258702721446753, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.90992126464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.92445068359376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16490, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.20044087022542953, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.011017902847379446, + "grad_norm": 0.12401829659938812, + "kept_token_count": 654.3859375, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9927991092205047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007200900558382272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5337677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5920379638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16500, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19957164973020552, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.014300204440951347, + "grad_norm": 0.1442761868238449, + "kept_token_count": 655.115625, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922450840473175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007754920888692141, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.99989318847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1813995361328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16510, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.200095833837986, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.012942564580589533, + "grad_norm": 0.1167827844619751, + "kept_token_count": 654.44375, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929220914840698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007077906373888254, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7541748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.45813903808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16520, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20005248188972474, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.013849902432411909, + "grad_norm": 0.1488659530878067, + "kept_token_count": 654.6234375, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9925766468048096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074233547784388065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.181103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5699981689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16530, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2003144934773445, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.012349765561521054, + "grad_norm": 0.09408094733953476, + "kept_token_count": 654.3828125, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9920958757400513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007904119789600372, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.0406005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0529083251953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.1993918925523758, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.013964294223114848, + "grad_norm": 0.09328773617744446, + "kept_token_count": 655.271875, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927758693695068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224120665341616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.79400939941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1716705322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16550, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19994536638259888, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.013436249177902936, + "grad_norm": 0.09562066942453384, + "kept_token_count": 654.7046875, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9926662504673004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007333756424486637, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.31507873535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0520843505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.20017682611942292, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.013953275419771672, + "grad_norm": 0.08410751074552536, + "kept_token_count": 654.625, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9921157717704773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007884235214442014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.7851135253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.37892456054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16570, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.20049112141132355, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.015036035794764757, + "grad_norm": 0.13580924272537231, + "kept_token_count": 654.81875, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924344718456268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007565540168434382, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7396179199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.262890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16580, + "valid_token_count": 819.025 + }, + { + "actual_token_mask_ratio": 0.20088794231414794, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.01449746172875166, + "grad_norm": 0.10071313381195068, + "kept_token_count": 653.69375, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927744925022125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007225522864609956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.25287475585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2428314208984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16590, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19978724122047425, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.016530222771689296, + "grad_norm": 0.08931830525398254, + "kept_token_count": 655.0625, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920365571975708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007963448110967875, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.70574951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3481719970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16600, + "valid_token_count": 818.6125 + }, + { + "actual_token_mask_ratio": 0.20097731947898864, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.013588658813387156, + "grad_norm": 0.14153467118740082, + "kept_token_count": 653.859375, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9923684060573578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007631598319858312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9377807617187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.29698486328124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16610, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.20022391080856322, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.01426814841106534, + "grad_norm": 0.10094969719648361, + "kept_token_count": 654.309375, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922173738479614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007782634068280459, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.07903442382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.505224609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16620, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20013669580221177, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.013354792166501284, + "grad_norm": 0.08616449683904648, + "kept_token_count": 654.646875, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992905056476593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007094951439648866, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.7488494873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.12644958496094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16630, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.19954935759305953, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.016308870818465947, + "grad_norm": 0.13846053183078766, + "kept_token_count": 655.1921875, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9922662258148194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007733762636780739, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.2142608642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2645568847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16640, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19997499883174896, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.013610842917114497, + "grad_norm": 0.14127086102962494, + "kept_token_count": 654.571875, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9924635589122772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007536452915519476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.2145141601562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7012023925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16650, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20035929083824158, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.014509098511189223, + "grad_norm": 0.12421752512454987, + "kept_token_count": 654.4265625, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927168786525726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00728312972933054, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.19256896972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0367401123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16660, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.19962289184331894, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.01620026221498847, + "grad_norm": 0.09627486020326614, + "kept_token_count": 654.971875, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920659780502319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007934030890464783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6119750976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.555029296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16670, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20079455971717836, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.01465204106643796, + "grad_norm": 0.13232895731925964, + "kept_token_count": 654.184375, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926068186759949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007393181696534157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.00599670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.04624938964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16680, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19961532205343246, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.017526782117784023, + "grad_norm": 0.10532815009355545, + "kept_token_count": 655.04375, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.992503958940506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007496043108403683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.67042541503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7786834716797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16690, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20051468461751937, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.01355586675927043, + "grad_norm": 0.10071521252393723, + "kept_token_count": 654.3796875, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.992687314748764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073126907460391525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.16819152832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.56248779296874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16700, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.1994367390871048, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.011897844541817904, + "grad_norm": 0.10500098764896393, + "kept_token_count": 655.1765625, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9927377820014953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007262230571359396, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6554656982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3556884765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16710, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.19959374219179155, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.012930057710036635, + "grad_norm": 0.1039661318063736, + "kept_token_count": 654.9796875, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9926096081733704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007390394061803818, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.21058044433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6913787841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16720, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.20028034448623658, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.013610359374433756, + "grad_norm": 0.1216711774468422, + "kept_token_count": 654.3734375, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9924121737480164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007587833516299725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4205261230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0490386962891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16730, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20003005415201186, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.012786479387432336, + "grad_norm": 0.09953834861516953, + "kept_token_count": 654.859375, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928844571113586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007115539256483316, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.68601379394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.75868835449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16740, + "valid_token_count": 818.609375 + }, + { + "actual_token_mask_ratio": 0.20038993656635284, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.015609614737331867, + "grad_norm": 0.10497555136680603, + "kept_token_count": 654.3890625, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.992845869064331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007154134474694729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8455505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.66395874023436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16750, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.1996936246752739, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.01311375917866826, + "grad_norm": 0.11889950931072235, + "kept_token_count": 655.2015625, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926905155181884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007309495098888874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.14096984863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0009704589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16760, + "valid_token_count": 818.6890625 + }, + { + "actual_token_mask_ratio": 0.19913723766803743, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.014288145676255226, + "grad_norm": 0.10237903892993927, + "kept_token_count": 655.7140625, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.992926687002182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070733225904405115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.08263244628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2187805175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16770, + "valid_token_count": 818.7546875 + }, + { + "actual_token_mask_ratio": 0.20097194910049437, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.013866876810789108, + "grad_norm": 0.11400524526834488, + "kept_token_count": 653.9703125, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9929099380970001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007090053986757994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4338623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5664794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16780, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.19991667419672013, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.013080741232261061, + "grad_norm": 0.07750622928142548, + "kept_token_count": 654.596875, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930461347103119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953856907784939, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5979309082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4962463378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16790, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1986928641796112, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.014730143919587135, + "grad_norm": 0.13499385118484497, + "kept_token_count": 655.8046875, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926416277885437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007358371932059527, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1100708007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2308380126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16800, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.20021491646766662, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.01403261125087738, + "grad_norm": 0.13816505670547485, + "kept_token_count": 654.415625, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9933721244335174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006627882085740566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.5296173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6608612060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19934694617986679, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.014373738877475261, + "grad_norm": 0.10497580468654633, + "kept_token_count": 655.0765625, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929799675941468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007020044419914484, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.8529998779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0312744140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16820, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.200316858291626, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.015720181073993443, + "grad_norm": 0.14401690661907196, + "kept_token_count": 654.30625, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.992640244960785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007359767891466617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0533874511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.42609252929685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16830, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20009581595659257, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.012407044135034084, + "grad_norm": 0.10959995537996292, + "kept_token_count": 654.8890625, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928004682064057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007199519034475088, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5948760986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1820861816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16840, + "valid_token_count": 818.7078125 + }, + { + "actual_token_mask_ratio": 0.2001907005906105, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.013448031712323427, + "grad_norm": 0.10242714732885361, + "kept_token_count": 654.5609375, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9926067233085633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073932777158916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.95323486328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4221984863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20040451884269714, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.0135640615131706, + "grad_norm": 0.13871943950653076, + "kept_token_count": 654.25, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9921861112117767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007813885435461998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.5959747314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.03404846191404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16860, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.19958941638469696, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.013839517952874303, + "grad_norm": 0.1277334839105606, + "kept_token_count": 655.00625, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9926763713359833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007323639746755361, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5700408935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33958740234374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16870, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.19971189349889756, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.012868545670062303, + "grad_norm": 0.09298762679100037, + "kept_token_count": 655.040625, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.992180061340332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007819940987974405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.08986511230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.37333984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16880, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20055678188800813, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.012490858789533377, + "grad_norm": 0.08803454041481018, + "kept_token_count": 654.33125, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9933212339878082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006678775046020746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.41441345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.07134704589845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16890, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.19982748329639435, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.014999301685020328, + "grad_norm": 0.10974973440170288, + "kept_token_count": 654.553125, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922865033149719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007713504321873188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3821746826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2148162841797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16900, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.2000893235206604, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.012223413586616516, + "grad_norm": 0.089283786714077, + "kept_token_count": 654.9203125, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934352695941925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006564735528081655, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 462.55706481933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9486602783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16910, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.19971684664487838, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.013445546105504035, + "grad_norm": 0.1284436285495758, + "kept_token_count": 654.890625, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9928671002388001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007132909167557955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2040740966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.51592712402345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16920, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.199912428855896, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.012982398830354213, + "grad_norm": 0.09347455203533173, + "kept_token_count": 654.790625, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9922365665435791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007763438764959574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5950164794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.73822631835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16930, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19925301671028137, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.012177057377994061, + "grad_norm": 0.11245238780975342, + "kept_token_count": 655.3, + "learning_rate": 1.30396481572432e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9924337506294251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007566240895539522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6107208251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4121398925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16940, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19999587088823317, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.012949258461594582, + "grad_norm": 0.11588748544454575, + "kept_token_count": 654.6640625, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.992515629529953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007484376151114702, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.7420288085938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2743316650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16950, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.19925307482481003, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.0113416139036417, + "grad_norm": 0.10912016779184341, + "kept_token_count": 655.228125, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9926826059818268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007317397277802229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.7737579345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8816436767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16960, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.1998407930135727, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.013108196016401053, + "grad_norm": 0.07411402463912964, + "kept_token_count": 654.91875, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9920962750911713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007903715781867503, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.03130798339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.96242370605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16970, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.20016162991523742, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.014142143446952105, + "grad_norm": 0.11220496147871017, + "kept_token_count": 654.34375, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9923607110977173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007639297004789114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2850738525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7140350341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16980, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.1991362154483795, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.01500446917489171, + "grad_norm": 0.10761523246765137, + "kept_token_count": 655.3796875, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9927334249019623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007266575098037719, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.50684509277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.18714599609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 16990, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.1997663974761963, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.014400131721049547, + "grad_norm": 0.11275367438793182, + "kept_token_count": 654.69375, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929684281349183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031562458723784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8268188476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.18800354003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17000, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19989453256130219, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.015333822742104531, + "grad_norm": 0.10820729285478592, + "kept_token_count": 654.5765625, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9919548034667969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008045203424990178, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3745574951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1530059814453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17010, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.2000660240650177, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.013539591245353222, + "grad_norm": 0.11170180886983871, + "kept_token_count": 654.615625, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9921637415885926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007836254686117173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8261199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.93833618164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17020, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.19991958141326904, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.012763595534488558, + "grad_norm": 0.09114310890436172, + "kept_token_count": 654.590625, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929356634616852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00706433542072773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8563934326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0210418701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17030, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19877877384424208, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.014165995083749295, + "grad_norm": 0.10661730915307999, + "kept_token_count": 655.51875, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9921395957469941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00786042883992195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.4791595458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8169799804688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17040, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.19929437786340715, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.013147678691893816, + "grad_norm": 0.1328786015510559, + "kept_token_count": 655.1375, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9925750315189361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007424966711550951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.987548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9574249267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17050, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2004440665245056, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014378172066062688, + "grad_norm": 0.09456652402877808, + "kept_token_count": 654.1046875, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9922975242137909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007702474761754274, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7546417236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3696594238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17060, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20026377886533736, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.01223777960985899, + "grad_norm": 0.09872591495513916, + "kept_token_count": 654.3546875, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.992152065038681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007847925554960966, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9298370361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1164245605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17070, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.19945085942745208, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.014284545183181762, + "grad_norm": 0.10871623456478119, + "kept_token_count": 655.134375, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9929540038108826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007046004850417375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1741638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9321685791016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19924877136945723, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.015222764015197754, + "grad_norm": 0.10501552373170853, + "kept_token_count": 655.121875, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9925630629062653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007436933647841215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0382110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6706298828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17090, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.19937436431646346, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.01276510814204812, + "grad_norm": 0.11128261685371399, + "kept_token_count": 654.9140625, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.992621773481369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00737822949886322, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.4046356201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.55787658691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17100, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.20029945969581603, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.01566800959408283, + "grad_norm": 0.10811319947242737, + "kept_token_count": 654.3, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9926234781742096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007376526016741991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2627746582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.31311950683596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17110, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2001437649130821, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.011875058757141232, + "grad_norm": 0.09545492380857468, + "kept_token_count": 654.6203125, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9929133176803588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007086681108921767, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1086761474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.128271484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17120, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20082247406244277, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.01432696869596839, + "grad_norm": 0.08143924176692963, + "kept_token_count": 654.04375, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929211735725403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007078814320266247, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4064605712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3596496582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17130, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20029266327619552, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015068263560533524, + "grad_norm": 0.10043251514434814, + "kept_token_count": 654.0484375, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9929744482040406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00702555226162076, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1973815917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.45325622558596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17140, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.19962930232286452, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.013084373809397221, + "grad_norm": 0.10003484785556793, + "kept_token_count": 654.6359375, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9926297247409821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007370278239250183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.4704223632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9905670166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17150, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.20016960203647613, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.013221439113840461, + "grad_norm": 0.07774010300636292, + "kept_token_count": 654.575, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930185616016388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006981428246945143, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5329528808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0516296386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17160, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.20020442754030227, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.013344888109713793, + "grad_norm": 0.12598642706871033, + "kept_token_count": 654.3234375, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928358793258667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007164140604436398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1946594238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62152709960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17170, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.20054040998220443, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.015515113063156605, + "grad_norm": 0.15342356264591217, + "kept_token_count": 654.2375, + "learning_rate": 1.264695080126559e-05, + "loss": 0.019, + "mgd_cos_sim": 0.992939043045044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007060958445072174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0899200439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3760192871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17180, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.19975951611995696, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013155233021825552, + "grad_norm": 0.10128382593393326, + "kept_token_count": 654.7578125, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929036617279052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007096336502581835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8543182373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.29400329589845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17190, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19913312941789627, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.013080053869634867, + "grad_norm": 0.09260821342468262, + "kept_token_count": 655.1234375, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9925192952156067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007480692584067583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.65684204101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4037231445312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17200, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20068268179893495, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.01267796941101551, + "grad_norm": 0.10275901854038239, + "kept_token_count": 654.190625, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9925261437892914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007473850063979626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4693145751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.25010070800784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17210, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.19891498535871505, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.015484080277383327, + "grad_norm": 0.11648383736610413, + "kept_token_count": 655.3390625, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927714705467224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007228522654622793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.13271179199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7859130859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17220, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20032197386026382, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.014864844270050526, + "grad_norm": 0.1090356707572937, + "kept_token_count": 654.2796875, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9926571309566498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007342871185392142, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.5387268066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6789245605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17230, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.20076631456613542, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.012819850584492087, + "grad_norm": 0.10723943263292313, + "kept_token_count": 654.3515625, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929257810115815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070742107927799225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4650115966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.55384826660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17240, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.19974396526813507, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.01429950874298811, + "grad_norm": 0.10965277999639511, + "kept_token_count": 654.853125, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9926521241664886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073478810489177706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8256530761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.34141540527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17250, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20028564631938933, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.015580682177096605, + "grad_norm": 0.12252683192491531, + "kept_token_count": 654.1890625, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9923181235790253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007681871391832828, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.0428161621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3604248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17260, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.2007619857788086, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.01268449006602168, + "grad_norm": 0.12853053212165833, + "kept_token_count": 654.1984375, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9925415933132171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00745841795578599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5447967529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6775299072266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17270, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.19930939376354218, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.013461307995021344, + "grad_norm": 0.08620021492242813, + "kept_token_count": 655.0640625, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9935078680515289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064921342767775055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7438568115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.180078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.1997242584824562, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.012948114797472954, + "grad_norm": 0.0793871060013771, + "kept_token_count": 654.403125, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9924125611782074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00758743304759264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.0682037353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7315734863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17290, + "valid_token_count": 817.721875 + }, + { + "actual_token_mask_ratio": 0.1996818721294403, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.013238303735852242, + "grad_norm": 0.08325628191232681, + "kept_token_count": 655.0234375, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9927419066429138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007258096523582935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.12726135253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4219024658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17300, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.1993776261806488, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.01497338991612196, + "grad_norm": 0.114521324634552, + "kept_token_count": 655.2453125, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9923215627670288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007678442262113094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.1906402587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3930969238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17310, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20009902566671373, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.014102133829146624, + "grad_norm": 0.10105740278959274, + "kept_token_count": 654.26875, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9926724910736084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00732750603929162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.8519683837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.62033996582034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17320, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.2000541865825653, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.015452256612479686, + "grad_norm": 0.10390353947877884, + "kept_token_count": 654.659375, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9922312557697296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007768749911338091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.48804626464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5157440185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17330, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.20080947428941726, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.011870114365592599, + "grad_norm": 0.09543212503194809, + "kept_token_count": 653.89375, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9924545466899872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007545453310012817, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8284576416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.33192443847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17340, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.1986660659313202, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.01328786863014102, + "grad_norm": 0.1012599766254425, + "kept_token_count": 655.6640625, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926856338977814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007314360700547695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.03997497558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8299102783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19980505853891373, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.01188144669868052, + "grad_norm": 0.1127682775259018, + "kept_token_count": 654.896875, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9928383231163025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00716167725622654, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6803436279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0104217529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17360, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20093046724796296, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.014344771485775709, + "grad_norm": 0.08673068135976791, + "kept_token_count": 653.9890625, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927642107009887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007235798239707947, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3268676757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.87201538085935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17370, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20075657814741135, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.012706825509667397, + "grad_norm": 0.11957256495952606, + "kept_token_count": 654.1859375, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9926126599311829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007387336529791355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.4903106689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7901672363281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17380, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.19969292134046554, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.013907164800912142, + "grad_norm": 0.096731036901474, + "kept_token_count": 654.6140625, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9926679372787476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007332073897123337, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.63047485351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9674041748047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17390, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.19929833561182023, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.01585535379126668, + "grad_norm": 0.12191502749919891, + "kept_token_count": 654.8984375, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.992885309457779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007114692963659764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.45384826660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5782073974609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17400, + "valid_token_count": 817.8984375 + }, + { + "actual_token_mask_ratio": 0.19987354576587676, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.013274727063253522, + "grad_norm": 0.08821727335453033, + "kept_token_count": 654.696875, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9926026582717895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007397347036749125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2646453857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9385498046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19888193607330323, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.01304775346070528, + "grad_norm": 0.1100541353225708, + "kept_token_count": 655.3046875, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9921321928501129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007867820095270872, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.34521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17420, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.20049077570438384, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.012932783830910921, + "grad_norm": 0.10849037021398544, + "kept_token_count": 654.4171875, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9926945865154266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073054124601185325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.0710205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.89882202148436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17430, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.20036722868680953, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.010877087153494358, + "grad_norm": 0.09527807682752609, + "kept_token_count": 654.421875, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9927614212036133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072385801933705805, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.84606018066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6377380371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17440, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20098651945590973, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.01134243574924767, + "grad_norm": 0.11598407477140427, + "kept_token_count": 653.7171875, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9932310283184052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006768963579088449, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.5722320556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2400848388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17450, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20004989802837372, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.015258902311325073, + "grad_norm": 0.13396309316158295, + "kept_token_count": 654.4265625, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9926906645298004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007309350185096264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.80602111816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3519805908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17460, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.19965683072805404, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.013992095552384853, + "grad_norm": 0.10412535816431046, + "kept_token_count": 655.0421875, + "learning_rate": 1.217466402591191e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9920792877674103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007920716144144534, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.49733581542966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6989715576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17470, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.20009442865848542, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.01398416915908456, + "grad_norm": 0.11313953250646591, + "kept_token_count": 654.396875, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9928135693073272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007186413276940584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.77727966308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.16921081542966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17480, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.1993149846792221, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.014622007496654987, + "grad_norm": 0.10605229437351227, + "kept_token_count": 655.1078125, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9927193462848664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072806629352271555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.97838134765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6622802734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17490, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19972283840179444, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.011952865775674582, + "grad_norm": 0.09297332167625427, + "kept_token_count": 654.7515625, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9929529547691345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070470418781042096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.57021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.66404724121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17500, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20055891275405885, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.012087015807628632, + "grad_norm": 0.117613285779953, + "kept_token_count": 653.790625, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9932022035121918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00679779639467597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3144195556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9903839111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17510, + "valid_token_count": 817.8109375 + }, + { + "actual_token_mask_ratio": 0.20062395632267, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.015944152139127254, + "grad_norm": 0.11241921782493591, + "kept_token_count": 654.3984375, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9922799170017242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077200720086693765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.98427429199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3296295166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17520, + "valid_token_count": 818.640625 + }, + { + "actual_token_mask_ratio": 0.19924333840608596, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.015872630570083857, + "grad_norm": 0.09933759272098541, + "kept_token_count": 655.1359375, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9929399251937866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007060074992477894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.7974822998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.06048889160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17530, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19995131343603134, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.015447022300213576, + "grad_norm": 0.1068827286362648, + "kept_token_count": 654.65625, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.992717570066452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007282438222318888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2495635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4252014160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17540, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.19923546761274338, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.013500622659921645, + "grad_norm": 0.13980895280838013, + "kept_token_count": 654.878125, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9929408550262451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007059139013290405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.73614807128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1555236816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17550, + "valid_token_count": 817.8140625 + }, + { + "actual_token_mask_ratio": 0.20036844462156295, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.01366705372929573, + "grad_norm": 0.08956310153007507, + "kept_token_count": 654.1390625, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9924683570861816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007531648874282837, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 452.0187103271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1413635253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17560, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20011795312166214, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.014605710422620178, + "grad_norm": 0.07908473908901215, + "kept_token_count": 654.775, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9927054107189178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007294600922614336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.3058837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8925354003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17570, + "valid_token_count": 818.5921875 + }, + { + "actual_token_mask_ratio": 0.19949401468038558, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.014620490092784166, + "grad_norm": 0.10192839056253433, + "kept_token_count": 654.740625, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926955878734589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073044121265411375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7775024414062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.65204467773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17580, + "valid_token_count": 817.9109375 + }, + { + "actual_token_mask_ratio": 0.1997772067785263, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.014750204328447581, + "grad_norm": 0.10966677218675613, + "kept_token_count": 654.74375, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928102433681488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007189769484102726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5182220458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.88544311523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17590, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2003824979066849, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.01355750085785985, + "grad_norm": 0.12449968606233597, + "kept_token_count": 654.190625, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926889955997467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007311009801924229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.3350372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.35565490722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17600, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.19919862747192382, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.012777629587799311, + "grad_norm": 0.09252553433179855, + "kept_token_count": 655.3828125, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9927900493144989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209946680814028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.68662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9133270263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17610, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20020660161972045, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.011491435207426548, + "grad_norm": 0.09152338653802872, + "kept_token_count": 654.55625, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9927118599414826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007288140151649714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.63463439941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.40228881835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19994204193353654, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.016573163215070963, + "grad_norm": 0.10889770835638046, + "kept_token_count": 654.88125, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9920820236206055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791797786951065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.5566131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.34132385253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17630, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.2002980813384056, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.014968256000429392, + "grad_norm": 0.11965866386890411, + "kept_token_count": 654.48125, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9928485691547394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007151427958160639, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8419952392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5386993408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17640, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20076885521411897, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.014184568868950009, + "grad_norm": 0.0808042660355568, + "kept_token_count": 653.9578125, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9928713083267212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007128690183162689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.65977783203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.55487060546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17650, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2009307324886322, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.014107593661174178, + "grad_norm": 0.07246133685112, + "kept_token_count": 654.153125, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9919570028781891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008042989764362574, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.59580993652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8339416503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17660, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.19954584985971452, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.011167802195996047, + "grad_norm": 0.09879729151725769, + "kept_token_count": 654.975, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9925682425498963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00743175195530057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3441131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.08994750976564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17670, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20071255713701247, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.012516214232891797, + "grad_norm": 0.11631886661052704, + "kept_token_count": 653.8859375, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9925406277179718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007459378242492676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.52781677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.90247802734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17680, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.20033937096595764, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.014783333148807287, + "grad_norm": 0.10174811631441116, + "kept_token_count": 654.528125, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9928448140621186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007155189756304026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4069519042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7386016845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17690, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.20000778436660765, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.01298255817964673, + "grad_norm": 0.09926068037748337, + "kept_token_count": 654.534375, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9926881194114685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007311865780502557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.4805023193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.2746307373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17700, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.20009839832782744, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.012396053411066532, + "grad_norm": 0.09523460268974304, + "kept_token_count": 654.3734375, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9924275159835816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007572470419108868, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.52357177734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7943939208984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17710, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.19974576085805892, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.01360276499763131, + "grad_norm": 0.10959089547395706, + "kept_token_count": 655.009375, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9927595496177674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007240455783903599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9414001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.95176391601564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17720, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20009246170520784, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.01297828946262598, + "grad_norm": 0.0883639007806778, + "kept_token_count": 654.6734375, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9930067121982574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006993291713297367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.42376098632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.07835083007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17730, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2001933678984642, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.012355205044150353, + "grad_norm": 0.10929988324642181, + "kept_token_count": 654.4546875, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928070425987243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00719295246526599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.22491760253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.65888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17740, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19959105402231217, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.013960700016468763, + "grad_norm": 0.12736767530441284, + "kept_token_count": 655.175, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9929642915725708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070357157848775385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.22705993652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.12353515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17750, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.20016274005174636, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.01158179584890604, + "grad_norm": 0.09910118579864502, + "kept_token_count": 654.575, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9926455438137054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007354457955807447, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.7420654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.38031616210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17760, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.1993864119052887, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.012906566262245178, + "grad_norm": 0.11887241154909134, + "kept_token_count": 655.0890625, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9925809979438782, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007419000752270222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0259307861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5123687744141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17770, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20010229349136352, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.01331157460808754, + "grad_norm": 0.14312170445919037, + "kept_token_count": 654.6109375, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9927159249782562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007284082006663084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5710479736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3135528564453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17780, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.19953390210866928, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.013333135098218919, + "grad_norm": 0.09047696739435196, + "kept_token_count": 654.89375, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928915321826934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00710847033187747, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.97991333007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3938018798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17790, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.20037952512502671, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.013200948061421514, + "grad_norm": 0.09168185293674469, + "kept_token_count": 654.4421875, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926939368247986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007306052837520838, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.4425903320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.57777404785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17800, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.2008819356560707, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.01404167115688324, + "grad_norm": 0.11405613273382187, + "kept_token_count": 653.8734375, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9929348170757294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00706517519429326, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.012548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6670776367188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17810, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.200141079723835, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.015250701922923326, + "grad_norm": 0.10987207293510437, + "kept_token_count": 654.26875, + "learning_rate": 1.160855154534595e-05, + "loss": 0.019, + "mgd_cos_sim": 0.992567366361618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007432641834020615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6243591308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62591247558595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17820, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.20014657825231552, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.013498644530773162, + "grad_norm": 0.16472013294696808, + "kept_token_count": 654.4765625, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9928295969963074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007170402072370052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4976745605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.823779296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17830, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20022906363010406, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.01423643296584487, + "grad_norm": 0.11210338771343231, + "kept_token_count": 654.440625, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.991834944486618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008165055606514215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3108306884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3843933105469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17840, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.2005662277340889, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.014293399732559919, + "grad_norm": 0.09598522633314133, + "kept_token_count": 653.8765625, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992778730392456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007221269793808461, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3889862060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5540496826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17850, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.20110995322465897, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.013338511437177658, + "grad_norm": 0.12319150567054749, + "kept_token_count": 653.8796875, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930018901824951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006998102366924286, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3458312988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.22259826660155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17860, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2004925027489662, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.014597578160464763, + "grad_norm": 0.10832693427801132, + "kept_token_count": 654.3890625, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928893446922302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071106554009020325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.80426025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17870, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.1992143362760544, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013235746324062348, + "grad_norm": 0.09488566219806671, + "kept_token_count": 655.1546875, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926166653633117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00738334022462368, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 470.2669403076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.705322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17880, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.20027063637971879, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013360368320718408, + "grad_norm": 0.10858052223920822, + "kept_token_count": 654.2984375, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9925380647182465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007461930997669697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.1493621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7062133789062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17890, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.20103243440389634, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012616213597357272, + "grad_norm": 0.0847695842385292, + "kept_token_count": 653.9484375, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9932235300540924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006776459980756044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.93273010253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.13555297851565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17900, + "valid_token_count": 818.49375 + }, + { + "actual_token_mask_ratio": 0.20073354095220566, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.012172298924997448, + "grad_norm": 0.08487315475940704, + "kept_token_count": 654.1390625, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9930466949939728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953298300504685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.92160949707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4881256103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17910, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19906761199235917, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.015600744169205427, + "grad_norm": 0.13128744065761566, + "kept_token_count": 654.996875, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9923690915107727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076309128664433954, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 468.03338928222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0530700683594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17920, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.1993196949362755, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.011976229399442673, + "grad_norm": 0.08248391002416611, + "kept_token_count": 655.09375, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9931133687496185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006886627711355686, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.92319946289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7520263671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17930, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.1995203360915184, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.01361685125157237, + "grad_norm": 0.10039777308702469, + "kept_token_count": 654.9171875, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9924198687076569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007580127753317356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.98472290039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5222900390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17940, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19974211901426314, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.012746039312332869, + "grad_norm": 0.10555869340896606, + "kept_token_count": 654.73125, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9924748301506042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007525177672505379, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.08008728027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0367462158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17950, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.1993884563446045, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.01262952173128724, + "grad_norm": 0.11983900517225266, + "kept_token_count": 655.1671875, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9922887146472931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007711296901106834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.14671936035154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23732604980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17960, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.20008083134889604, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.015157311921939254, + "grad_norm": 0.09647977352142334, + "kept_token_count": 654.4671875, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922642886638642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007735717389732599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.25067138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8577056884766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1996690332889557, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.011009094025939702, + "grad_norm": 0.10132110863924026, + "kept_token_count": 654.95625, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9929266452789307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007073351927101612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.96164855957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8323059082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17980, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20034219920635224, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.01171663417480886, + "grad_norm": 0.11939477175474167, + "kept_token_count": 654.4609375, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9930428087711334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069571900181472305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.18068237304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8498901367187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 17990, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.19938733875751496, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.01330766398459673, + "grad_norm": 0.1337268352508545, + "kept_token_count": 655.4484375, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928410947322845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007158900704234838, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1830322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.81679382324216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18000, + "valid_token_count": 818.6828125 + }, + { + "actual_token_mask_ratio": 0.1996401369571686, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.012533331895247102, + "grad_norm": 0.1265329122543335, + "kept_token_count": 654.853125, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.992867362499237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007132632564753294, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9923950195313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8457427978516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2007777839899063, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.012769363261759282, + "grad_norm": 0.08856622129678726, + "kept_token_count": 654.0296875, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9927728474140167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007227154914289713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8509155273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4853485107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18020, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20051529556512832, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.013239051774144172, + "grad_norm": 0.10994940996170044, + "kept_token_count": 654.28125, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926395535469055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073604494333267215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.5033752441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8147674560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18030, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.19948560148477554, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.013544482551515103, + "grad_norm": 0.0941028892993927, + "kept_token_count": 654.9015625, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9931891858577728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006810818612575531, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.6300964355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5399169921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18040, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.20021476000547409, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.013115565199404955, + "grad_norm": 0.09064565598964691, + "kept_token_count": 654.4328125, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9933063983917236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00669359490275383, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.3291381835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.24248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18050, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20028720051050186, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.012852580193430186, + "grad_norm": 0.12080638855695724, + "kept_token_count": 654.6109375, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9925352275371552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007464748434722423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.55535583496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.02288513183595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18060, + "valid_token_count": 818.5609375 + }, + { + "actual_token_mask_ratio": 0.2001611515879631, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.013101901300251484, + "grad_norm": 0.12314771115779877, + "kept_token_count": 654.5375, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9928908705711365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071091210469603535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.62383422851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3697174072266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18070, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20130489021539688, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.013977378886193038, + "grad_norm": 0.1067592054605484, + "kept_token_count": 653.53125, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9923111975193024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00768879298120737, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.60418090820315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5227508544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18080, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2001931384205818, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.014463572856038808, + "grad_norm": 0.10937386006116867, + "kept_token_count": 654.515625, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9926422834396362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007357722986489535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2576599121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.49986572265624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18090, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.19930373579263688, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.012519350461661815, + "grad_norm": 0.08699485659599304, + "kept_token_count": 655.1078125, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.992529833316803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007470151223242283, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9764709472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7809997558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18100, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.20045313090085984, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.013097234163433313, + "grad_norm": 0.11681873351335526, + "kept_token_count": 654.1515625, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9923165678977967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007683438342064619, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.29930725097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.171484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18110, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.19919579327106476, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.012924744514748454, + "grad_norm": 0.10214876383543015, + "kept_token_count": 655.3375, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9926142215728759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007385790813714266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2751892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8081420898437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19995700865983962, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.012641625152900815, + "grad_norm": 0.10446152836084366, + "kept_token_count": 654.7859375, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9921684026718139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0078315868973732, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9260223388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.813916015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18130, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19994972944259642, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.016040511336177586, + "grad_norm": 0.09731155633926392, + "kept_token_count": 654.63125, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9931259095668793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006874097138643264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7145660400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.25859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18140, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19978512823581696, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.012003267370164394, + "grad_norm": 0.10072607547044754, + "kept_token_count": 654.6671875, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933959186077118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006604071799665689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1863586425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9136993408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18150, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20083799362182617, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.01236240854486823, + "grad_norm": 0.08035144209861755, + "kept_token_count": 653.7234375, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9926616966724395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007338302209973335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0152557373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2490570068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18160, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.19985350966453552, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.0159673516638577, + "grad_norm": 0.07254593819379807, + "kept_token_count": 654.7234375, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9927783727645874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007221628911793232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6430908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0084503173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18170, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19886343479156493, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.013503234274685383, + "grad_norm": 0.07805519551038742, + "kept_token_count": 655.4703125, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9925823092460633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00741768479347229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.71026611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.08449096679686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18180, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.1993849754333496, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.01356500261463225, + "grad_norm": 0.11742491275072098, + "kept_token_count": 654.8640625, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9932180106639862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006781987100839615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6772491455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.78367614746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18190, + "valid_token_count": 817.95625 + }, + { + "actual_token_mask_ratio": 0.2003227010369301, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.01168974321335554, + "grad_norm": 0.08141624182462692, + "kept_token_count": 654.6375, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9928858757019043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00711410641670227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2895751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5488525390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18200, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20007588118314742, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.012237206101417542, + "grad_norm": 0.12474405020475388, + "kept_token_count": 654.5484375, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9927503049373627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007249699532985687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.67413330078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0566040039063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18210, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2002914994955063, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.015829675365239382, + "grad_norm": 0.12288400530815125, + "kept_token_count": 654.5109375, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9925867319107056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074132663197815415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9496215820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9806274414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18220, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19913727194070815, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.012814477598294616, + "grad_norm": 0.09892921894788742, + "kept_token_count": 655.0515625, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9927141010761261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007285892870277167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2431396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.66260681152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18230, + "valid_token_count": 817.928125 + }, + { + "actual_token_mask_ratio": 0.20005051642656327, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.012867083307355642, + "grad_norm": 0.10373857617378235, + "kept_token_count": 654.3265625, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9927645087242126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007235495559871197, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.75461730957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.04790649414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18240, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19920834600925447, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.013186618685722351, + "grad_norm": 0.10173400491476059, + "kept_token_count": 655.046875, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930130898952484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006986917182803154, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2659973144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.226171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18250, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.1997242897748947, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.013378428854048252, + "grad_norm": 0.11161167174577713, + "kept_token_count": 654.7953125, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9924324810504913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007567516528069973, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.74512939453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.704638671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18260, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.1998716861009598, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.01501568970270455, + "grad_norm": 0.08916211873292923, + "kept_token_count": 654.7109375, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9922395944595337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077604098245501515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3818420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.31476440429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18270, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20086973756551743, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.012779860850423574, + "grad_norm": 0.10498036444187164, + "kept_token_count": 653.94375, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928250670433044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007174933701753617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2772613525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9472259521484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18280, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.20058115273714067, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.013540349761024117, + "grad_norm": 0.09768389165401459, + "kept_token_count": 653.89375, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9925824880599976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007417517621070147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.12242736816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.48939819335936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18290, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.20002285689115523, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.012531953305006028, + "grad_norm": 0.11853455752134323, + "kept_token_count": 654.5125, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9929412722587585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007058734726160765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.23695068359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2034515380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18300, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.200344255566597, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.014518971275538205, + "grad_norm": 0.09847686439752579, + "kept_token_count": 654.3453125, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9926334321498871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007366572134196758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4059997558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.713330078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.19966886639595033, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.012391322385519743, + "grad_norm": 0.084252268075943, + "kept_token_count": 654.9078125, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9924378633499146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075621319003403185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.79392395019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.35216064453124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18320, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19946344196796417, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.01618980970233679, + "grad_norm": 0.10818485915660858, + "kept_token_count": 654.925, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9922700762748718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007729927636682987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.00860290527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9356719970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18330, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.19994163811206817, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.013344764476642013, + "grad_norm": 0.10942340642213821, + "kept_token_count": 654.8390625, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.992187750339508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007812254410237074, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7381622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7998413085937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18340, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.199688820540905, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.011474285926669835, + "grad_norm": 0.09980782121419907, + "kept_token_count": 654.85, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.993320882320404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006679094489663839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9819641113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2716949462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18350, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.19979336857795715, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.012825526995584369, + "grad_norm": 0.1382213979959488, + "kept_token_count": 654.8484375, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929075360298156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092462945729494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8688690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.51637878417966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18360, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1996248796582222, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01300687501206994, + "grad_norm": 0.12449265271425247, + "kept_token_count": 654.859375, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.992901474237442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007098537031561136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.56963500976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0957366943359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18370, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.19965219795703887, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.015196643210947514, + "grad_norm": 0.11109447479248047, + "kept_token_count": 654.99375, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9929390192031861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007060984242707491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.3347991943359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.64765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18380, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.19980385154485703, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.011800586292520166, + "grad_norm": 0.0945332795381546, + "kept_token_count": 655.015625, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9928683876991272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007131610531359911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.275732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.07732543945315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18390, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.20075109750032424, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.014635010808706283, + "grad_norm": 0.15655218064785004, + "kept_token_count": 654.0703125, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9925655782222748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007434406504034996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0922424316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.04546813964845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18400, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.19990580081939696, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.014696385897696018, + "grad_norm": 0.11697836965322495, + "kept_token_count": 655.115625, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9927206575870514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007279346790164709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0632385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8361389160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18410, + "valid_token_count": 818.8 + }, + { + "actual_token_mask_ratio": 0.2007855638861656, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.011957243736833333, + "grad_norm": 0.14904925227165222, + "kept_token_count": 653.990625, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933098793029785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006690114177763462, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.4228820800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6124542236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18420, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20002980381250382, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.014406357239931823, + "grad_norm": 0.11739691346883774, + "kept_token_count": 654.6734375, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9928963541984558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007103640865534544, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9038391113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4643890380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18430, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.2007426917552948, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.01219306867569685, + "grad_norm": 0.09794113785028458, + "kept_token_count": 654.15, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9922331631183624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007766840420663357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.66404724121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0358551025391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18440, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.19959644824266434, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.012639267230406403, + "grad_norm": 0.12559561431407928, + "kept_token_count": 655.0140625, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9928319752216339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007168031204491854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0903564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7447845458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18450, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.20020339488983155, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.014125501411035657, + "grad_norm": 0.08228235691785812, + "kept_token_count": 654.4265625, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.992946594953537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007053420972079038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8471405029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9650573730469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18460, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2002401426434517, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.012375509180128575, + "grad_norm": 0.099154032766819, + "kept_token_count": 654.4265625, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.99262375831604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007376242429018021, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8117248535156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18470, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.19948953241109849, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.012425740528851748, + "grad_norm": 0.08872848749160767, + "kept_token_count": 655.2578125, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9929559111595154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00704409098252654, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.3548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5884246826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18480, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19936499744653702, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.01287214457988739, + "grad_norm": 0.09926696121692657, + "kept_token_count": 655.08125, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9924644529819489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007535554375499487, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4852294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.92682189941405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18490, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.1997189074754715, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.013223234377801419, + "grad_norm": 0.0879148617386818, + "kept_token_count": 654.915625, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931203126907349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006879683304578066, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.09593200683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1776641845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.2002849742770195, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.014610608294606208, + "grad_norm": 0.10906218737363815, + "kept_token_count": 654.55625, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927889406681061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072110495530068874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.3598297119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7073516845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18510, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.2005001500248909, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.014336004294455052, + "grad_norm": 0.12249267101287842, + "kept_token_count": 654.4984375, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928396165370941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071603888645768166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.64610595703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4781921386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18520, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.20046979039907456, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.011706243641674519, + "grad_norm": 0.10742075741291046, + "kept_token_count": 654.28125, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9931860387325286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068139548413455485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.8476501464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2882110595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18530, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.20011839121580124, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.013727391604334116, + "grad_norm": 0.11819186061620712, + "kept_token_count": 654.690625, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9924704134464264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007529591675847769, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4885528564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.192822265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18540, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.20099201798439026, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.012941778032109141, + "grad_norm": 0.10360515862703323, + "kept_token_count": 653.8140625, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9930685520172119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006931445747613907, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.25962829589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94654846191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18550, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.19895529001951218, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.013478460628539324, + "grad_norm": 0.10847204923629761, + "kept_token_count": 655.2609375, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9928829073905945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007117089256644249, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.84157409667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.17578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18560, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.20012654215097428, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.014335976261645556, + "grad_norm": 0.11303311586380005, + "kept_token_count": 654.1515625, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924137830734253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007586215995252133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.1260040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8211242675781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18570, + "valid_token_count": 817.8171875 + }, + { + "actual_token_mask_ratio": 0.20070704072713852, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.011842485330998898, + "grad_norm": 0.09322411566972733, + "kept_token_count": 654.2515625, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.992473840713501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007526155561208725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0817413330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2849822998047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18580, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.20060402452945708, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.014096410572528839, + "grad_norm": 0.10505110025405884, + "kept_token_count": 654.225, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927993237972259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00720066549256444, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4696319580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5395965576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18590, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20019969791173936, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.012917667813599109, + "grad_norm": 0.14389657974243164, + "kept_token_count": 654.253125, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9931642472743988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006835771072655916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.57615661621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.17938537597655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18600, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.19886520951986314, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.01437479886226356, + "grad_norm": 0.10251118987798691, + "kept_token_count": 655.75, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929240465164184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007075957767665386, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.198388671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.14248046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18610, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.2004685178399086, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.012876030337065459, + "grad_norm": 0.13228857517242432, + "kept_token_count": 654.0359375, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928846597671509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007115333713591099, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3587982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.56964721679685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18620, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.20018952041864396, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.01279202289879322, + "grad_norm": 0.08548706769943237, + "kept_token_count": 654.6453125, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928305864334106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007169407233595848, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.1781768798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9008544921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18630, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.20011912137269974, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.012352509144693612, + "grad_norm": 0.0813354104757309, + "kept_token_count": 654.775, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9925487101078033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007451293990015984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1696044921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.29310913085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18640, + "valid_token_count": 818.5875 + }, + { + "actual_token_mask_ratio": 0.20006667971611022, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.011648645531386137, + "grad_norm": 0.1178627461194992, + "kept_token_count": 654.571875, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9930469036102295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069531072862446305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.92287292480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.12558898925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2004546657204628, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.012668237648904323, + "grad_norm": 0.11318767815828323, + "kept_token_count": 654.3578125, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9928814828395843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007118519861251116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.9211883544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0347015380859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18660, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.20010514557361603, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.010212215222418308, + "grad_norm": 0.10777457803487778, + "kept_token_count": 654.4875, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9927658021450043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072341937571764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5763793945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1278350830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18670, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.1992174431681633, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.013262257445603609, + "grad_norm": 0.11367996782064438, + "kept_token_count": 655.59375, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9929993152618408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007000686507672072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5722137451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.33546142578126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18680, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.19971537142992019, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.01349142538383603, + "grad_norm": 0.10642798244953156, + "kept_token_count": 654.5921875, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9930187165737152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069812658242881295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.79200134277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5299987792969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18690, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20053299367427826, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.014659861754626035, + "grad_norm": 0.09300877898931503, + "kept_token_count": 654.1046875, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9925211608409882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007478834129869938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.49158630371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.27943115234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18700, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.19987669438123704, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.013182990159839392, + "grad_norm": 0.11420869082212448, + "kept_token_count": 654.81875, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.992612886428833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073871172964572905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7846954345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.37145080566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18710, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19919484853744507, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.01412165593355894, + "grad_norm": 0.10683116316795349, + "kept_token_count": 655.059375, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927231252193451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007276874128729105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.70517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5412078857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18720, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.1997198596596718, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.011871888907626272, + "grad_norm": 0.10329750925302505, + "kept_token_count": 655.0125, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9919302284717559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008069772832095623, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8923736572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.09307250976565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18730, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19953736215829848, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.015838573314249517, + "grad_norm": 0.12108606845140457, + "kept_token_count": 655.421875, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9925547301769256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007445293013006449, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9197235107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.05287475585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18740, + "valid_token_count": 818.8015625 + }, + { + "actual_token_mask_ratio": 0.20032233148813247, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.013054842036217451, + "grad_norm": 0.09410084038972855, + "kept_token_count": 654.3859375, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.993394935131073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00660506309941411, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9623107910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.17539672851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18750, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.19928280413150787, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.013387763034552335, + "grad_norm": 0.10583440959453583, + "kept_token_count": 655.2546875, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9928221940994263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007177822291851044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0546630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6303375244141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18760, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19919798970222474, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.012947619054466486, + "grad_norm": 0.08927496522665024, + "kept_token_count": 655.35625, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9932300209999084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006769962422549724, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.68764343261716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.479150390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18770, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19982802122831345, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.012532639829441906, + "grad_norm": 0.08973363041877747, + "kept_token_count": 654.8421875, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.016, + "mgd_cos_sim": 0.993061226606369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006938772089779377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.02040710449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.152783203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18780, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.19997515231370927, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.01198805053718388, + "grad_norm": 0.10207638889551163, + "kept_token_count": 654.521875, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9926603853702545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00733961770310998, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6128662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.09373168945314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18790, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19992109537124633, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.014581514522433281, + "grad_norm": 0.1113143339753151, + "kept_token_count": 655.040625, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9922659277915955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077340899035334585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0517944335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3800048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18800, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.20011652261018753, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.012178447656333446, + "grad_norm": 0.12609831988811493, + "kept_token_count": 654.5859375, + "learning_rate": 1.003734268121676e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9923178613185882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007682143710553646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2637145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.20250244140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18810, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.200179760158062, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.013525561802089215, + "grad_norm": 0.1246936097741127, + "kept_token_count": 654.696875, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9931378960609436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068621054291725155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.41216735839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2258728027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18820, + "valid_token_count": 818.5546875 + }, + { + "actual_token_mask_ratio": 0.19966262429952622, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.013052884815260768, + "grad_norm": 0.1024252250790596, + "kept_token_count": 654.746875, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9932295620441437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006770438421517611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.73744812011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5491180419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18830, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.19989251345396042, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.013411087449640036, + "grad_norm": 0.07579734921455383, + "kept_token_count": 654.6765625, + "learning_rate": 9.990559236068628e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9928929448127747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007107060682028532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.4082427978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3293151855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18840, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.20003669261932372, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.014151268266141414, + "grad_norm": 0.12317118793725967, + "kept_token_count": 654.715625, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9926938652992249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007306129019707441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.96958923339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8243713378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18850, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.20019788891077042, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.012430333346128464, + "grad_norm": 0.08784221112728119, + "kept_token_count": 654.253125, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9925434947013855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007456495892256498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.20200500488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1446594238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18860, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19962052404880523, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.012663541454821825, + "grad_norm": 0.1272978037595749, + "kept_token_count": 655.015625, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931118547916412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006888151355087757, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.16054382324216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.0369873046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18870, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.19979008734226228, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.012015316914767028, + "grad_norm": 0.08686954528093338, + "kept_token_count": 654.8921875, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9926810443401337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007318958081305027, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.73216552734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.88828735351564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18880, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.19992245733737946, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.013013607216998934, + "grad_norm": 0.09362667798995972, + "kept_token_count": 654.4640625, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9933046460151672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066953436471521854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.79111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1958801269531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18890, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.1999418169260025, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.012419191654771566, + "grad_norm": 0.07913322746753693, + "kept_token_count": 654.7015625, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9927383542060852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261640671640634, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.8474151611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.36228332519534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18900, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.19909567683935164, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.011818305961787701, + "grad_norm": 0.08984430134296417, + "kept_token_count": 655.340625, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9929981887340545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007001830637454987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.11329956054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1026550292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18910, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20056369453668593, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.014230981655418872, + "grad_norm": 0.12532372772693634, + "kept_token_count": 654.215625, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9928154170513153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007184582483023405, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.02251586914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5413116455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18920, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20021844804286956, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.013949592784047126, + "grad_norm": 0.10694063454866409, + "kept_token_count": 654.4109375, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9921970129013061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00780298225581646, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.59986572265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.41481323242186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18930, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.19872553646564484, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.011836006958037614, + "grad_norm": 0.13001097738742828, + "kept_token_count": 655.721875, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.992533129453659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007466867938637734, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.64195861816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7634216308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18940, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20024501532316208, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.013169385166838765, + "grad_norm": 0.10815185308456421, + "kept_token_count": 654.440625, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9925194263458252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074805647134780886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7415710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6272674560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18950, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.20002079159021377, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.012688825186342002, + "grad_norm": 0.10137815028429031, + "kept_token_count": 654.5640625, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9924156963825226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00758429691195488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5160614013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9746887207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18960, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.1988583743572235, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.01516607846133411, + "grad_norm": 0.07509541511535645, + "kept_token_count": 655.30625, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9927757918834687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224193681031466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.9658477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4595428466797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18970, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.1999310225248337, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.013892565248534083, + "grad_norm": 0.10686036944389343, + "kept_token_count": 654.6625, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9929604828357697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007039526849985123, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.5011291503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7668151855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18980, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20016978532075883, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.013583853747695685, + "grad_norm": 0.15102048218250275, + "kept_token_count": 654.6359375, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9924266099929809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007573394104838371, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.51670837402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2656677246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 18990, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2001319408416748, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.012602290557697415, + "grad_norm": 0.18656004965305328, + "kept_token_count": 654.5484375, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9921941936016083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007805795967578888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5872833251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83408508300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19000, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.2000760927796364, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.012631787732243539, + "grad_norm": 0.10167212039232254, + "kept_token_count": 654.54375, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9928637206554413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007136269193142653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.10255432128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2650939941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19010, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20021897405385972, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.012284375540912152, + "grad_norm": 0.08492107689380646, + "kept_token_count": 654.2796875, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9928302884101867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007169714849442244, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8803985595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.79827575683595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19020, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19998595863580704, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.014702435024082661, + "grad_norm": 0.1190822497010231, + "kept_token_count": 654.6984375, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9930397748947144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00696021169424057, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.67662353515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1686645507813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19030, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19917906820774078, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.015410208608955145, + "grad_norm": 0.10701125860214233, + "kept_token_count": 655.565625, + "learning_rate": 9.68009423081259e-06, + "loss": 0.019, + "mgd_cos_sim": 0.9927576661109925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007242345996201038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.226416015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5110565185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19040, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.19964821338653566, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.013843365013599396, + "grad_norm": 0.09158919751644135, + "kept_token_count": 654.7375, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9924277603626251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007572229206562042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9754211425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.96259155273435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19050, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20107657015323638, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.015578749869018794, + "grad_norm": 0.10295980423688889, + "kept_token_count": 653.6703125, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0192, + "mgd_cos_sim": 0.992834460735321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007165541965514421, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.03327026367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3479522705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19060, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.1998710662126541, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.01389675228856504, + "grad_norm": 0.12414810061454773, + "kept_token_count": 654.853125, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.9926430940628052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007356901280581951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.52512817382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7358917236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19070, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20083512216806412, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.013779774028807878, + "grad_norm": 0.10250212252140045, + "kept_token_count": 653.93125, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9925788104534149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007421186007559299, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.63837280273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.81296691894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19080, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.19959818720817565, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.012684532906860113, + "grad_norm": 0.11794350296258926, + "kept_token_count": 654.803125, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9929054260253907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007094580866396427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.7597351074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5154327392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19090, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20063145607709884, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.01418867353349924, + "grad_norm": 0.14076218008995056, + "kept_token_count": 654.1125, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9928843915462494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071156035177409645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.14048156738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.36351623535154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19100, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19954922646284104, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.01471354765817523, + "grad_norm": 0.09865240007638931, + "kept_token_count": 655.228125, + "learning_rate": 9.57203463328011e-06, + "loss": 0.018, + "mgd_cos_sim": 0.993356853723526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006643145810812712, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9694519042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7590637207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19110, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.19964829981327056, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.014777943212538957, + "grad_norm": 0.1238405704498291, + "kept_token_count": 654.60625, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9926577031612396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007342289295047521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9306182861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9548614501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19120, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.20035054236650468, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.013852651696652174, + "grad_norm": 0.14333575963974, + "kept_token_count": 654.64375, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9926881551742553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007311836630105972, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8199096679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7755615234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19130, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.2006665736436844, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.014284207206219434, + "grad_norm": 0.12048787623643875, + "kept_token_count": 653.865625, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9922655701637269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007734435983002186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.93337097167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0967742919922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19140, + "valid_token_count": 818.0109375 + }, + { + "actual_token_mask_ratio": 0.20006602108478547, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.013804733380675316, + "grad_norm": 0.1027471274137497, + "kept_token_count": 654.521875, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9929116725921631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070883195847272875, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7537902832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7742095947266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19150, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19989503175020218, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.014248184766620398, + "grad_norm": 0.09923820197582245, + "kept_token_count": 654.64375, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0181, + "mgd_cos_sim": 0.9923909187316895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00760909141972661, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2976928710938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.25562438964846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19160, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.1995898589491844, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.010859658056870104, + "grad_norm": 0.084695003926754, + "kept_token_count": 655.2953125, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9929705917835235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007029417157173157, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1156494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3827880859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19170, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.2000439614057541, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.014224856533110142, + "grad_norm": 0.09647604823112488, + "kept_token_count": 654.671875, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9930941820144653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069058328866958615, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7696502685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1230102539063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19180, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.19961638301610946, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.013496664259582758, + "grad_norm": 0.0849965512752533, + "kept_token_count": 655.0921875, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9926582455635071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007341765332967043, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.563623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0157836914062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19190, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20021891295909883, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.01489926059730351, + "grad_norm": 0.11258640140295029, + "kept_token_count": 654.321875, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9929651856422425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007034820783883333, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2644775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.06585083007815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19200, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.20020825862884523, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.012071967590600252, + "grad_norm": 0.09867803007364273, + "kept_token_count": 654.1890625, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9928409636020661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007159049063920975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.30861511230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4587066650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19210, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.20122900158166884, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.013300797529518604, + "grad_norm": 0.08874272555112839, + "kept_token_count": 653.725, + "learning_rate": 9.402882113991094e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9926074922084809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007392492052167654, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.4047607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.12726135253905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19220, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.20123230665922165, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.013963202852755785, + "grad_norm": 0.1287999451160431, + "kept_token_count": 653.2609375, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9924460113048553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007553989626467228, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.57646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.567041015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19230, + "valid_token_count": 817.8375 + }, + { + "actual_token_mask_ratio": 0.20051705688238144, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.01285524400882423, + "grad_norm": 0.09258196502923965, + "kept_token_count": 653.9765625, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932029485702515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006797041837126016, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.3849243164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.32521362304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19240, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.20074819326400756, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.014311229716986418, + "grad_norm": 0.11766255646944046, + "kept_token_count": 654.0984375, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9932009041309356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006799111142754555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.4420501708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7627288818359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19250, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.1992812782526016, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.012800437025725841, + "grad_norm": 0.08500675112009048, + "kept_token_count": 655.2515625, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9931629717350006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006837030034512281, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.32769470214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6480194091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19260, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19963512867689132, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.01067827818915248, + "grad_norm": 0.12645822763442993, + "kept_token_count": 655.16875, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9931633472442627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006836635619401931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.35723571777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0151824951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19270, + "valid_token_count": 818.584375 + }, + { + "actual_token_mask_ratio": 0.20122349113225937, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.016444517858326436, + "grad_norm": 0.11723264306783676, + "kept_token_count": 653.25625, + "learning_rate": 9.310962368239861e-06, + "loss": 0.02, + "mgd_cos_sim": 0.9929808080196381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070191888138651844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.66767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7391418457031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19280, + "valid_token_count": 817.8203125 + }, + { + "actual_token_mask_ratio": 0.20100050419569016, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.011699146032333374, + "grad_norm": 0.10996080935001373, + "kept_token_count": 653.6921875, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9925838828086853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074161218479275705, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.79486389160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7206207275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19290, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19966761022806168, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.014530988968908786, + "grad_norm": 0.10732875764369965, + "kept_token_count": 654.8234375, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9922289252281189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00777106611058116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7541564941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23204650878904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19300, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.19984334260225295, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.013381561869755388, + "grad_norm": 0.09100174158811569, + "kept_token_count": 654.909375, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9928854048252106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007114590797573328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.479541015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8380126953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.1998932421207428, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.014798360690474511, + "grad_norm": 0.10553918033838272, + "kept_token_count": 654.6171875, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9928253650665283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007174635026603937, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1195922851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.53655395507815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19320, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19979910254478456, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.012160199135541916, + "grad_norm": 0.09416421502828598, + "kept_token_count": 654.865625, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993513423204422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006486586388200521, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.8330780029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1547607421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19330, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20021657645702362, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.012543929554522038, + "grad_norm": 0.1304280310869217, + "kept_token_count": 654.478125, + "learning_rate": 9.21929147942456e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9931168615818023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006883146427571774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6564483642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5054077148437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19340, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19872932881116867, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.011420175293460488, + "grad_norm": 0.07536041736602783, + "kept_token_count": 655.628125, + "learning_rate": 9.204037474105611e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9928483009338379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007151701487600803, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0212799072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9131164550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20001300126314164, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.012701347563415765, + "grad_norm": 0.09552524238824844, + "kept_token_count": 654.528125, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.992706972360611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007293032947927713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.83095397949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3811370849609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19360, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.1996395155787468, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.014809220843017102, + "grad_norm": 0.1216871440410614, + "kept_token_count": 654.8671875, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0183, + "mgd_cos_sim": 0.9929653406143188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00703466972336173, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.026513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7673797607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19370, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.19794345051050186, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.012468606792390346, + "grad_norm": 0.08751019090414047, + "kept_token_count": 656.3828125, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9927347302436829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007265283353626728, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3340606689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.06946716308596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19380, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2002764329314232, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.013814467005431652, + "grad_norm": 0.12473341822624207, + "kept_token_count": 654.5609375, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9928711593151093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071288345381617544, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5633911132812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9744567871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19390, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.20007974207401275, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.014064878132194281, + "grad_norm": 0.09525571763515472, + "kept_token_count": 654.6875, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9931581854820252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006841801106929779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0848907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.19871520996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19400, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.20068167597055436, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.012955918535590171, + "grad_norm": 0.1327500194311142, + "kept_token_count": 653.6859375, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9935082972049714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00649169972166419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.12127685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0879241943359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19410, + "valid_token_count": 817.803125 + }, + { + "actual_token_mask_ratio": 0.199879752099514, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.013510757591575384, + "grad_norm": 0.12098928540945053, + "kept_token_count": 654.43125, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9927032887935638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007296713907271624, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5030181884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1094116210937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19420, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.19978259801864623, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.011434897454455495, + "grad_norm": 0.13962696492671967, + "kept_token_count": 654.559375, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9934178590774536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065821451134979725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.49403991699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9115814208984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19430, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.19902715235948562, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.016058764699846507, + "grad_norm": 0.11913127452135086, + "kept_token_count": 655.3359375, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0199, + "mgd_cos_sim": 0.9923677980899811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007632210943847894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.51511535644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9125244140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19440, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2017360061407089, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.009813312999904155, + "grad_norm": 0.1459348350763321, + "kept_token_count": 653.2421875, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9930460393428803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953964103013277, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.9110809326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.46185913085935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19450, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.20059140622615815, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.0121237781830132, + "grad_norm": 0.08241380751132965, + "kept_token_count": 653.8421875, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9931705594062805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006829453445971012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.41375427246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.83558044433596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19460, + "valid_token_count": 817.90625 + }, + { + "actual_token_mask_ratio": 0.20039480030536652, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.01213960158638656, + "grad_norm": 0.11776167899370193, + "kept_token_count": 654.2078125, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9929795742034913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070204133167862896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8293060302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.40191955566405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19470, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19994909316301346, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.011785949254408479, + "grad_norm": 0.09718801081180573, + "kept_token_count": 654.5140625, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9932418882846832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006758111156523227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8739044189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2637115478516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.19929119795560837, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.011984108528122307, + "grad_norm": 0.07486991584300995, + "kept_token_count": 655.5109375, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9929571568965911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007042843382805586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.30625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.55804138183595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19490, + "valid_token_count": 818.6609375 + }, + { + "actual_token_mask_ratio": 0.19967133849859237, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.011882195388898254, + "grad_norm": 0.07430844008922577, + "kept_token_count": 654.88125, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9926606893539429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007339302264153957, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1937225341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.82575073242185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19500, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.1992386743426323, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.016149085946381092, + "grad_norm": 0.09441252797842026, + "kept_token_count": 655.159375, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0196, + "mgd_cos_sim": 0.9931598246097565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006840169988572597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.62038269042966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2164611816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19510, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.20078794956207274, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.012752919644117355, + "grad_norm": 0.11645182222127914, + "kept_token_count": 653.753125, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9927890956401825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072109119966626166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.10608825683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3980651855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19520, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.20078830569982528, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.015088716195896268, + "grad_norm": 0.10593731701374054, + "kept_token_count": 653.959375, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0191, + "mgd_cos_sim": 0.9920564532279968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007943545561283826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6667022705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7989562988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19530, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19907797127962112, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.012037565000355244, + "grad_norm": 0.0925525575876236, + "kept_token_count": 655.4984375, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9923321843147278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007667831517755986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5136444091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7220123291016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19540, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2004040837287903, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.013832566048949956, + "grad_norm": 0.10945995897054672, + "kept_token_count": 654.225, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9930835902690888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006916401442140341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.78905029296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23460693359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19550, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19976067095994948, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.012636264506727456, + "grad_norm": 0.09557498246431351, + "kept_token_count": 654.9640625, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9927606344223022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007239361666142941, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.5264617919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83018188476564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19560, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.199570731818676, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.01349994782358408, + "grad_norm": 0.09371481835842133, + "kept_token_count": 655.178125, + "learning_rate": 8.870263878954767e-06, + "loss": 0.017, + "mgd_cos_sim": 0.993036299943924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963692419230938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.44850158691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.63983459472655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19570, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.20135581493377686, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.01245673792436719, + "grad_norm": 0.09828345477581024, + "kept_token_count": 653.4390625, + "learning_rate": 8.855176526188428e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9928528428077698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007147146295756102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.5629852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5736511230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19580, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.1993689090013504, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011434105224907398, + "grad_norm": 0.09577113389968872, + "kept_token_count": 655.3625, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9932304561138153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006769539136439562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.60882873535155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.24749145507815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19590, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.19975198805332184, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.01303750891238451, + "grad_norm": 0.11582250148057938, + "kept_token_count": 654.9640625, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.992989307641983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007010691799223423, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0027221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8129577636719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19600, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.19988800585269928, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.012507458124309779, + "grad_norm": 0.08466129004955292, + "kept_token_count": 654.7984375, + "learning_rate": 8.809959340413891e-06, + "loss": 0.016, + "mgd_cos_sim": 0.993088710308075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069112972356379036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.11375732421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1593963623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19610, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2003091022372246, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.012394662713631988, + "grad_norm": 0.10254765301942825, + "kept_token_count": 654.2703125, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932655274868012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006734470650553703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.3987579345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9035888671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19620, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.20020875632762908, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.01383856227621436, + "grad_norm": 0.11111254245042801, + "kept_token_count": 654.5140625, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9931339681148529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006866031978279352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0623413085938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7780517578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19630, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1998799189925194, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.012393610086292028, + "grad_norm": 0.10290296375751495, + "kept_token_count": 654.70625, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9923534333705902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007646567840129137, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.0166961669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.17980041503904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19640, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2009609118103981, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.012990751676261425, + "grad_norm": 0.10380968451499939, + "kept_token_count": 653.75625, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9930095851421357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069904222153127195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.49896545410155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.8605682373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19650, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.19979409873485565, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.015255281608551741, + "grad_norm": 0.10020442306995392, + "kept_token_count": 654.715625, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9927370071411132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007262986339628697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.19452209472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.4673187255859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19660, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20018057525157928, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.015997742116451264, + "grad_norm": 0.11960765719413757, + "kept_token_count": 654.5953125, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0197, + "mgd_cos_sim": 0.9926311612129212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007368838787078858, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.27989501953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2347930908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19670, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.20007017254829407, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.01242821137420833, + "grad_norm": 0.10327364504337311, + "kept_token_count": 654.359375, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9930389881134033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006961017567664385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.60491943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.90758361816404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19680, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.19974290281534196, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.01250938167795539, + "grad_norm": 0.09784621745347977, + "kept_token_count": 654.971875, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9925248861312866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007475108653306961, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.65602722167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 387.29942932128904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19690, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.19911640882492065, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.012247668206691742, + "grad_norm": 0.10674034059047699, + "kept_token_count": 655.296875, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9930969953536988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006903001759201288, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.45, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1177093505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19700, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.19968414306640625, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.010800181468948723, + "grad_norm": 0.0768580287694931, + "kept_token_count": 654.7859375, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.992861670255661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007138330861926079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 454.1094543457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3723419189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19710, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.19980479925870895, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.012572567444294691, + "grad_norm": 0.09540735930204391, + "kept_token_count": 654.9921875, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9935040056705475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006495984457433224, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.82703552246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.87599487304686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19720, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19931090772151946, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.012003546487540007, + "grad_norm": 0.07777201384305954, + "kept_token_count": 654.9359375, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9933718323707581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066281726583838465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.6911895751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.67754821777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19730, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.19914367496967317, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.01265379535034299, + "grad_norm": 0.10963088274002075, + "kept_token_count": 655.046875, + "learning_rate": 8.614806769544507e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932532727718353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006746730208396912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.68349304199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0843780517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19740, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.20024475753307341, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.012779348483309149, + "grad_norm": 0.08953908085823059, + "kept_token_count": 654.4125, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9927386403083801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261356245726347, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.0152313232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.44382629394534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19750, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19951267242431642, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.014937463402748107, + "grad_norm": 0.10443729907274246, + "kept_token_count": 654.9203125, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0187, + "mgd_cos_sim": 0.992500513792038, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007499482482671738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5210876464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8596557617187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19760, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.20068995356559755, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.011585169844329357, + "grad_norm": 0.104945108294487, + "kept_token_count": 653.9875, + "learning_rate": 8.569956310579001e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9932179927825928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006782004795968533, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.41465759277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.32990112304685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19770, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.20008333176374435, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.011843099677935242, + "grad_norm": 0.09920048713684082, + "kept_token_count": 654.459375, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933065831661224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006693412642925977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.68658142089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.83453369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19780, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20041164606809617, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.01418042415753007, + "grad_norm": 0.07773303985595703, + "kept_token_count": 654.4140625, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9927748262882232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007225182186812162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.513134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7267150878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19790, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19945637583732606, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.014048458077013493, + "grad_norm": 0.0824771597981453, + "kept_token_count": 655.1203125, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.993210369348526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00678962841629982, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.40684204101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0501434326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19800, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.20017934441566468, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.012418269086629153, + "grad_norm": 0.07908888161182404, + "kept_token_count": 654.5640625, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9934737741947174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006526232417672873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.2523651123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.78566284179686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19810, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.1998078405857086, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.011088124942034483, + "grad_norm": 0.08991540223360062, + "kept_token_count": 654.6953125, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9929721295833588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00702787833288312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7816925048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8263366699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19820, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20060660541057587, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.013512295298278331, + "grad_norm": 0.11337839812040329, + "kept_token_count": 654.1296875, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9931559503078461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068440593779087065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.4194671630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7500274658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19830, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19839045256376267, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013085236400365829, + "grad_norm": 0.11828345060348511, + "kept_token_count": 655.671875, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9928919553756714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007108063716441393, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.92194213867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9375885009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19840, + "valid_token_count": 817.9453125 + }, + { + "actual_token_mask_ratio": 0.1998981222510338, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.01236018566414714, + "grad_norm": 0.09977764636278152, + "kept_token_count": 654.7171875, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9928564131259918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007143579330295325, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.5369049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.89947204589845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20010163486003876, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.01269336137920618, + "grad_norm": 0.11821290105581284, + "kept_token_count": 654.4984375, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9930732846260071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006926714070141315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.6336364746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9834838867188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20047377347946166, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.012294463673606515, + "grad_norm": 0.09187112748622894, + "kept_token_count": 654.228125, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9932718515396118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006728157587349415, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9716064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2524719238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19870, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1998004913330078, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.013739042263478041, + "grad_norm": 0.11287546157836914, + "kept_token_count": 654.9625, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9927223563194275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007277644984424114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.97344360351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.20909729003904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19880, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.200118051469326, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.012748288456350565, + "grad_norm": 0.10486824065446854, + "kept_token_count": 654.384375, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933636009693145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006636388599872589, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.34791870117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.68348693847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19890, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.1997861698269844, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.011610975302755833, + "grad_norm": 0.09109941124916077, + "kept_token_count": 654.5125, + "learning_rate": 8.376423166251816e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9931836903095246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006816321425139904, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.13056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.86517639160155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19900, + "valid_token_count": 817.9171875 + }, + { + "actual_token_mask_ratio": 0.20004526227712632, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.014982435945421457, + "grad_norm": 0.1020909994840622, + "kept_token_count": 654.4359375, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0191, + "mgd_cos_sim": 0.9917858123779297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008214183803647756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4113739013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.75328369140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.1995159089565277, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.01303562130779028, + "grad_norm": 0.11166664212942123, + "kept_token_count": 654.7734375, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9927886545658111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007211346924304962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6706604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 386.3232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19920, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20131878703832626, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.01754984902217984, + "grad_norm": 0.13198530673980713, + "kept_token_count": 653.534375, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0213, + "mgd_cos_sim": 0.9925332844257355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007466713897883892, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.7292785644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.83612670898435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19930, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.19887581765651702, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.011975081078708172, + "grad_norm": 0.08946570754051208, + "kept_token_count": 655.621875, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9929815649986267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070184431038796905, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.743310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.67291259765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19940, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.19998491406440735, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.012868114467710256, + "grad_norm": 0.11033303290605545, + "kept_token_count": 654.4171875, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9935473024845123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006452689319849015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.04783020019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8877349853516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19950, + "valid_token_count": 818.0046875 + }, + { + "actual_token_mask_ratio": 0.20024569183588029, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.015637533413246273, + "grad_norm": 0.12773923575878143, + "kept_token_count": 654.3984375, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0191, + "mgd_cos_sim": 0.9930591523647309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006940842699259519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5646270751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0339324951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19960, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.20074177384376526, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.011351788276806473, + "grad_norm": 0.11865268647670746, + "kept_token_count": 653.9765625, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936220228672028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006377969216555357, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6929656982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.34336242675784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19970, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19929994493722916, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.012479531299322843, + "grad_norm": 0.09285114705562592, + "kept_token_count": 655.003125, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9927099287509918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007290078978985548, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.68178405761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.64076538085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19980, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.20020000040531158, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.011061344994232059, + "grad_norm": 0.08851154893636703, + "kept_token_count": 654.3890625, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9928397238254547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071602812968194485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.78668823242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0933349609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 19990, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.20001614093780518, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.013281761016696692, + "grad_norm": 0.0948379635810852, + "kept_token_count": 654.4140625, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9929685056209564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031483668833971, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5923797607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.98549499511716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20000, + "valid_token_count": 818.0359375 + }, + { + "actual_token_mask_ratio": 0.19924102127552032, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.010918720765039325, + "grad_norm": 0.1169879361987114, + "kept_token_count": 655.2, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9925631642341614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074368506669998165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.394091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.63881225585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20010, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.20015119910240173, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.01235339166596532, + "grad_norm": 0.10535965859889984, + "kept_token_count": 654.3921875, + "learning_rate": 8.198983507064469e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9926897883415222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007310210727155209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 452.26218566894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18348388671876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20020, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.19985234290361403, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.0142026973888278, + "grad_norm": 0.10955052822828293, + "kept_token_count": 654.6578125, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9930175542831421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006982453260570764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.6675537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1338409423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20030, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19939843863248824, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.012479644920676946, + "grad_norm": 0.08539474010467529, + "kept_token_count": 655.0796875, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.993131297826767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006868718378245831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.5938446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5943878173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20040, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.199935582280159, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.010334063461050392, + "grad_norm": 0.08723133057355881, + "kept_token_count": 654.753125, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9931319177150726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006868091877549887, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.78224182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7260986328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20050, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2005433604121208, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.01123521295376122, + "grad_norm": 0.09504596889019012, + "kept_token_count": 654.1171875, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9927791118621826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007220900617539883, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.8558776855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.03420104980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20060, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19964466691017152, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.012887870008125902, + "grad_norm": 0.0938490629196167, + "kept_token_count": 654.8015625, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9932799577713013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672003123909235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4463623046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3670928955078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20070, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20080768167972565, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.013275961671024562, + "grad_norm": 0.1538187712430954, + "kept_token_count": 653.8140625, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.993007218837738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006992795690894127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2688354492187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.21734313964845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20080, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2001264661550522, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.012272090651094913, + "grad_norm": 0.06646344065666199, + "kept_token_count": 654.621875, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9934687197208405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531279440969229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.14613342285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3455413818359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20090, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2001842588186264, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.01206133645027876, + "grad_norm": 0.09571679681539536, + "kept_token_count": 654.5453125, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930571615695953, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006942846812307835, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.0186706542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9310272216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20100, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19998279660940171, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.014140819059684872, + "grad_norm": 0.11022260785102844, + "kept_token_count": 654.6296875, + "learning_rate": 8.066682997585045e-06, + "loss": 0.018, + "mgd_cos_sim": 0.992380040884018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007619963213801384, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.57634887695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15318298339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20110, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20027672201395036, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.012057663779705763, + "grad_norm": 0.11005578935146332, + "kept_token_count": 654.4625, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9929417431354522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007058258634060621, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.37371826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2943176269531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.20030322968959807, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.012596166133880616, + "grad_norm": 0.11352211236953735, + "kept_token_count": 654.2265625, + "learning_rate": 8.037375097441574e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9931091070175171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006890882086008787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.85097351074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.49119262695314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20130, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20067797750234603, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.012435691431164742, + "grad_norm": 0.1033903956413269, + "kept_token_count": 654.028125, + "learning_rate": 8.022733828838124e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9929535031318665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070464987307786945, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5248138427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68455200195314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20140, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.20032286942005156, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.012108912831172347, + "grad_norm": 0.08939153701066971, + "kept_token_count": 654.2046875, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9930282950401306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069717028178274635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1410766601563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.35405883789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20150, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20068527907133102, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.013547581527382136, + "grad_norm": 0.09152203798294067, + "kept_token_count": 653.9109375, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9928234636783599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007176535204052925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5976257324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.700390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20160, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.20078401118516923, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.013664236152544617, + "grad_norm": 0.10317646712064743, + "kept_token_count": 654.2625, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9930730521678924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006926937308162451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.10422668457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.72679443359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20170, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.20026578158140182, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.01116647282615304, + "grad_norm": 0.07863492518663406, + "kept_token_count": 654.334375, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9929829120635987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007017076015472412, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5559356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.10985107421874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20180, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19995132833719254, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.012986134178936482, + "grad_norm": 0.09213463217020035, + "kept_token_count": 654.7484375, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9925489485263824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074510638602077964, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.33885803222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58306884765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20190, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.20028768628835678, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.012015372142195701, + "grad_norm": 0.1290494203567505, + "kept_token_count": 654.1609375, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930221378803253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006977882888168097, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8251098632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4818481445312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20200, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.20089282244443893, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.01649254113435745, + "grad_norm": 0.11768987029790878, + "kept_token_count": 653.8140625, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0203, + "mgd_cos_sim": 0.9924154698848724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007584534026682377, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5204833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.09605102539064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20210, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.20109746158123015, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.013790776580572128, + "grad_norm": 0.10391858965158463, + "kept_token_count": 653.8359375, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9928121507167816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187825348228216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.300341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3517547607422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20220, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19974696785211563, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.01266662385314703, + "grad_norm": 0.10724998265504837, + "kept_token_count": 654.790625, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9927191972732544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007280801236629486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7447906494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6212921142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20230, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20121336579322815, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.012579190731048583, + "grad_norm": 0.12011735886335373, + "kept_token_count": 653.5875, + "learning_rate": 7.876790365120447e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932105481624603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067894491367042065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1045776367188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5073303222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20240, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.2003549039363861, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.01173413610085845, + "grad_norm": 0.09552516788244247, + "kept_token_count": 654.065625, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931101381778717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889871321618557, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6061065673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4400207519531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20250, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20070848017930984, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.012902427092194557, + "grad_norm": 0.09838379174470901, + "kept_token_count": 653.971875, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932846784591675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006715319305658341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.68567810058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5241394042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20260, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.2000923916697502, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.012574573140591383, + "grad_norm": 0.16197523474693298, + "kept_token_count": 654.5796875, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9926749646663666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007325027324259281, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.08058471679686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.03070678710935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20270, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.2000019669532776, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.011919006425887346, + "grad_norm": 0.09867909550666809, + "kept_token_count": 654.5875, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9930967628955841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069032390601933, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.361962890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.36549987792966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20280, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19980028718709947, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.013171754218637944, + "grad_norm": 0.08233966678380966, + "kept_token_count": 654.7640625, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931134104728698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006886586081236601, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.38917236328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.33424377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20290, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19970755726099015, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.014043038943782449, + "grad_norm": 0.058589495718479156, + "kept_token_count": 654.871875, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9926417350769043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007358252722769976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6720825195313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4598419189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20300, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19958197623491286, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.01308626625686884, + "grad_norm": 0.1122155711054802, + "kept_token_count": 655.0109375, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9927906453609466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209350541234017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.41943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2105346679688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.20017754137516022, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.01139589687809348, + "grad_norm": 0.09118065983057022, + "kept_token_count": 654.58125, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9931681632995606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006831842567771673, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.57648010253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8487548828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2006268322467804, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.01204120721668005, + "grad_norm": 0.07866779714822769, + "kept_token_count": 654.2921875, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9933545112609863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006645494792610407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.92333068847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.93483276367186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20330, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.1997946634888649, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.012745828367769718, + "grad_norm": 0.09607405960559845, + "kept_token_count": 654.9640625, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9930446147918701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006955377943813801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3797973632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8785369873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20340, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.2002866804599762, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.012828125292435289, + "grad_norm": 0.10139606148004532, + "kept_token_count": 654.3421875, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9925019085407257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007498097140341997, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.5671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.48387451171874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20350, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19917369931936263, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.012231743428856134, + "grad_norm": 0.092147596180439, + "kept_token_count": 655.259375, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9935595214366912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006440469436347485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.59601440429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20360, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20116260200738906, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.01336252298206091, + "grad_norm": 0.10425224900245667, + "kept_token_count": 653.3203125, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933135807514191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00668641934171319, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8156799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83208618164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20370, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.19926870465278626, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.015309058409184218, + "grad_norm": 0.09278994798660278, + "kept_token_count": 655.2265625, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9929082751274109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007091720495373011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.35825805664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5808074951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20380, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19936786741018295, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.010919422982260585, + "grad_norm": 0.09951958805322647, + "kept_token_count": 655.0734375, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9932088971138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791105028241873, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9063293457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4499877929687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20390, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.19992148727178574, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.013023260748013854, + "grad_norm": 0.10833890736103058, + "kept_token_count": 654.5765625, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9931809723377227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006819011364132166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.89307556152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.03810424804686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20400, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.20030345469713212, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.01099186111241579, + "grad_norm": 0.09016599506139755, + "kept_token_count": 654.3234375, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9929567337036133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007043275143951177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2160217285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.94336853027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20410, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.19931897521018982, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.012227426189929247, + "grad_norm": 0.11608748883008957, + "kept_token_count": 655.265625, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931296944618225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068703082390129564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2585418701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.96953735351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20420, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19994501024484634, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.010837336257100105, + "grad_norm": 0.0836063101887703, + "kept_token_count": 654.55625, + "learning_rate": 7.601896886850632e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936417937278748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006358199659734965, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.56578369140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9091857910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20430, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.19887536019086838, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.01307577001862228, + "grad_norm": 0.11421802639961243, + "kept_token_count": 655.4140625, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9931627690792084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006837231665849685, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0163299560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.48435363769534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20440, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19951419085264205, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.012297406513243913, + "grad_norm": 0.08232997357845306, + "kept_token_count": 655.0890625, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931190311908722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006880966294556856, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8065460205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3975311279297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20450, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.1996399536728859, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.01113742939196527, + "grad_norm": 0.09551694989204407, + "kept_token_count": 654.7015625, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9930777609348297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922239437699318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.90370178222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.607568359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20460, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.19861914664506913, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.011959382612258197, + "grad_norm": 0.10250771045684814, + "kept_token_count": 655.4890625, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9925784051418305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007421595416963101, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.58265991210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.78771057128904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20470, + "valid_token_count": 817.9515625 + }, + { + "actual_token_mask_ratio": 0.19985097795724868, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.015530095901340246, + "grad_norm": 0.10392466187477112, + "kept_token_count": 654.715625, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0188, + "mgd_cos_sim": 0.9934801757335663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006519827526062727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7226898193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.95283203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20480, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.19985251277685165, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.012415768578648567, + "grad_norm": 0.10735882818698883, + "kept_token_count": 654.859375, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9926474928855896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007352510932832956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.008349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8682495117188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20490, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.2003540113568306, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.010099503211677074, + "grad_norm": 0.09393354505300522, + "kept_token_count": 654.190625, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9930634915828704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006936517264693976, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7462951660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.17491149902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20500, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.1994380012154579, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.012614044174551965, + "grad_norm": 0.12390343844890594, + "kept_token_count": 654.8265625, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9927703380584717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007229676656424999, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3381378173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94085998535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20510, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.20034410804510117, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.010687489621341228, + "grad_norm": 0.08513031154870987, + "kept_token_count": 654.3125, + "learning_rate": 7.472809478761672e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933757185935974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006624278519302607, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.4999603271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0312805175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2000385746359825, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013063666690140963, + "grad_norm": 0.09767287969589233, + "kept_token_count": 654.490625, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9930552899837494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069447198882699014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.32541809082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.01189575195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20530, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2011660471558571, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.012873549666255712, + "grad_norm": 0.10343215614557266, + "kept_token_count": 653.6359375, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9926616549491882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007338330801576376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.99659729003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3188720703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20540, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.19961283057928086, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.012502799136564135, + "grad_norm": 0.15833839774131775, + "kept_token_count": 655.2796875, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934004127979279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006599600613117218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6741607666016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20550, + "valid_token_count": 818.70625 + }, + { + "actual_token_mask_ratio": 0.1998698055744171, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.011471007950603962, + "grad_norm": 0.0947311520576477, + "kept_token_count": 654.690625, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.992428320646286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007571667246520519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.16396179199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4151977539062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20560, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.19945164769887924, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.01314022745937109, + "grad_norm": 0.12907718122005463, + "kept_token_count": 655.175, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9930374324321747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006962561886757613, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2009613037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8593688964844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20570, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.1999428614974022, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.01384541681036353, + "grad_norm": 0.0968250185251236, + "kept_token_count": 654.5453125, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9927718579769135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007228136900812388, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3282775878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2245147705078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20580, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.19966383874416352, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.011782165570184588, + "grad_norm": 0.08551670610904694, + "kept_token_count": 654.6171875, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9938780725002289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006121914833784103, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.68052368164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7942260742187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20590, + "valid_token_count": 817.9265625 + }, + { + "actual_token_mask_ratio": 0.19998180419206618, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.01252956474199891, + "grad_norm": 0.11584848165512085, + "kept_token_count": 654.5171875, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9932441115379333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006755898240953684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1783477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1927459716797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20600, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.19877548664808273, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.012726529501378536, + "grad_norm": 0.08752557635307312, + "kept_token_count": 655.7171875, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.99279825091362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007201759051531554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.02931518554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.51563415527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20610, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.20106695592403412, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.011782951978966593, + "grad_norm": 0.10314787924289703, + "kept_token_count": 653.8515625, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931494355201721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006850569881498814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8405395507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6047821044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20620, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.19941513538360595, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.013729765871539712, + "grad_norm": 0.09430020302534103, + "kept_token_count": 655.0484375, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9928922295570374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007107779011130333, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.89385375976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2998291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20630, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20109318643808366, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.01239205915480852, + "grad_norm": 0.14124999940395355, + "kept_token_count": 653.5859375, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9931558132171631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006844202149659395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1627624511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4292877197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.19999074190855026, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.011264376156032086, + "grad_norm": 0.08972901850938797, + "kept_token_count": 654.521875, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931145250797272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006885460950434208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8677581787109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9694305419922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20650, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20089224874973297, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.01244291039183736, + "grad_norm": 0.11686807125806808, + "kept_token_count": 654.171875, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9921957433223725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007804262731224298, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1265502929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.20263671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20660, + "valid_token_count": 818.628125 + }, + { + "actual_token_mask_ratio": 0.19946549385786055, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.011085651023313403, + "grad_norm": 0.06924968957901001, + "kept_token_count": 655.325, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9930718123912812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006928186863660812, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 466.7243957519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4315887451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20670, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.19979631155729294, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.013490530615672468, + "grad_norm": 0.09524054825305939, + "kept_token_count": 654.5015625, + "learning_rate": 7.245158076724757e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9929040431976318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070959610864520075, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.5875732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.92290649414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20680, + "valid_token_count": 817.91875 + }, + { + "actual_token_mask_ratio": 0.20067844092845916, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.012921390775591134, + "grad_norm": 0.10763265937566757, + "kept_token_count": 654.3078125, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9929685056209564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031496427953243, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7750579833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2707183837891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20690, + "valid_token_count": 818.5703125 + }, + { + "actual_token_mask_ratio": 0.19976140558719635, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014128036005422473, + "grad_norm": 0.09352173656225204, + "kept_token_count": 654.8515625, + "learning_rate": 7.216869814524323e-06, + "loss": 0.018, + "mgd_cos_sim": 0.9923516750335694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007648327853530645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8937957763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1495300292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20700, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.20023203939199447, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.011743053374812008, + "grad_norm": 0.12036345154047012, + "kept_token_count": 654.4703125, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9929544806480408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007045519817620516, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2394104003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.057080078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20710, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20079250037670135, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.014564571250230075, + "grad_norm": 0.09559227526187897, + "kept_token_count": 654.0296875, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9932317614555359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067682497203350065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.99871826171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.26580200195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20720, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20045520067214967, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.012817875202745199, + "grad_norm": 0.07002856582403183, + "kept_token_count": 654.4078125, + "learning_rate": 7.174508393396057e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935562372207641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443762220442295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.72255554199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2337341308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20730, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.20037181973457335, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.011427969112992287, + "grad_norm": 0.11094389855861664, + "kept_token_count": 654.3140625, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9935330450534821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006466950848698616, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.27037353515624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2654693603516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20740, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20046872794628143, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.011919186171144246, + "grad_norm": 0.10457635670900345, + "kept_token_count": 654.4203125, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9935327231884002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006467266101390123, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.0500915527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.09649047851565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20750, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19945466220378877, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.01045940313488245, + "grad_norm": 0.13207054138183594, + "kept_token_count": 654.9921875, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9927423655986786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007257630676031112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.77434997558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.31968994140624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20760, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19970546960830687, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.012554635247215628, + "grad_norm": 0.10454697906970978, + "kept_token_count": 654.9203125, + "learning_rate": 7.118159700327839e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930148303508759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069851700216531755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.02403564453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1631713867188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20770, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.20048969686031343, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.012558046355843544, + "grad_norm": 0.10341596603393555, + "kept_token_count": 654.5296875, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9929717361927033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070282603614032265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4822021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22337951660154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20780, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.19972999393939972, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.011917732842266559, + "grad_norm": 0.09253250062465668, + "kept_token_count": 654.6546875, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9933030664920807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006696908734738827, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1774383544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9369354248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20790, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.19955404996871948, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.010492236306890845, + "grad_norm": 0.0988989919424057, + "kept_token_count": 654.8265625, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9932791888713837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006720815785229206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.4693176269531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4102752685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20800, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.20004866719245912, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.014228368643671274, + "grad_norm": 0.10313178598880768, + "kept_token_count": 654.71875, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9926097214221954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007390273734927177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.20098571777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.96058349609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20810, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.2005019411444664, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.01246692668646574, + "grad_norm": 0.10605702549219131, + "kept_token_count": 654.028125, + "learning_rate": 7.047939448292442e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9929810047149659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007018999103456736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.18785705566404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5608764648438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20820, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.19977981001138687, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.014280469249933958, + "grad_norm": 0.09150546789169312, + "kept_token_count": 654.9328125, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9928056538105011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007194341812282801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4442169189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.40053405761716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20830, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20040101259946824, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.01104139033704996, + "grad_norm": 0.11051448434591293, + "kept_token_count": 654.2796875, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.993301272392273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006698716804385186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.9405212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.0528228759766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20840, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20020890086889268, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.012502314848825335, + "grad_norm": 0.11712852120399475, + "kept_token_count": 654.409375, + "learning_rate": 7.005923110616313e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930310010910034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006969004590064287, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.2802337646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.20453491210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20850, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.20008271038532258, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.01243517268449068, + "grad_norm": 0.14002592861652374, + "kept_token_count": 654.3984375, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9935754597187042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006424534879624843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8361846923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0437286376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20860, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.19897826313972472, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.012434559082612396, + "grad_norm": 0.08948054164648056, + "kept_token_count": 655.471875, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931670129299164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00683298883959651, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.473876953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1473388671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20870, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.19874011129140853, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.011050567077472807, + "grad_norm": 0.09684311598539352, + "kept_token_count": 655.5640625, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935562610626221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443750578910112, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.5277099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.51459045410155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20880, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.19993042200803757, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.01279007876291871, + "grad_norm": 0.18454609811306, + "kept_token_count": 654.778125, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.993025541305542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006974457763135433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.5787017822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.83355102539065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20890, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.20020749568939208, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.01372060040012002, + "grad_norm": 0.09560474008321762, + "kept_token_count": 654.6859375, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9929390251636505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00706099160015583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2640045166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22243347167966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20900, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.20012191981077193, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.01325108939781785, + "grad_norm": 0.08724132925271988, + "kept_token_count": 654.5296875, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9929707288742066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007029267773032188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.43711547851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.38812561035155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20910, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.1991624727845192, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.013564938772469759, + "grad_norm": 0.08783319592475891, + "kept_token_count": 655.3390625, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9932886779308319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006711324490606785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.8725952148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5345886230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20920, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19941697716712953, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.011862058937549592, + "grad_norm": 0.3796194791793823, + "kept_token_count": 655.39375, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9930363357067108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963667180389166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.64585876464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4906280517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20930, + "valid_token_count": 818.6484375 + }, + { + "actual_token_mask_ratio": 0.2002918913960457, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.013189155980944634, + "grad_norm": 0.12477865815162659, + "kept_token_count": 654.3625, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9930063724517822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006993625313043594, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.273681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2031005859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20940, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.1996619373559952, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.012691094540059566, + "grad_norm": 0.09419766813516617, + "kept_token_count": 654.74375, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931963443756103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006803655624389648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3180633544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0824432373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20950, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20032687336206437, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.013663075119256973, + "grad_norm": 0.10444962233304977, + "kept_token_count": 654.31875, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9931020438671112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006897955015301704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.13341064453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.28587951660154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20960, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1995745673775673, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.013731562625616789, + "grad_norm": 0.1554388552904129, + "kept_token_count": 655.015625, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9926989436149597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007301059830933809, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 454.21422424316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8697998046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20970, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19984746277332305, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.013496720977127553, + "grad_norm": 0.1137022003531456, + "kept_token_count": 654.9296875, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.9928651690483093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007134831789880991, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.2282318115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.964208984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20980, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19904694259166716, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.012787324329838156, + "grad_norm": 0.11391797661781311, + "kept_token_count": 655.2375, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9934666097164154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533392146229744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.52740173339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.27527770996096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 20990, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.20009755194187165, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.011112062307074667, + "grad_norm": 0.06948257982730865, + "kept_token_count": 654.465625, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933479368686676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006652055401355028, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8299835205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.40000915527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21000, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2001506969332695, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.015035759285092353, + "grad_norm": 0.11069348454475403, + "kept_token_count": 654.5234375, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0184, + "mgd_cos_sim": 0.9932383716106414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006761628855019808, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.8753601074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4635925292969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.20027885288000108, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.01354797100648284, + "grad_norm": 0.09516787528991699, + "kept_token_count": 654.1265625, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9923739612102509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076260224916040896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2817626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0795166015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21020, + "valid_token_count": 817.9484375 + }, + { + "actual_token_mask_ratio": 0.20108335316181183, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.012052133865654468, + "grad_norm": 0.10784690082073212, + "kept_token_count": 653.9140625, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9929750025272369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007024999428540468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.92572326660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.24643859863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21030, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.200442373752594, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.013803793955594302, + "grad_norm": 0.09097015857696533, + "kept_token_count": 654.2265625, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9930375099182129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006962492223829031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5884155273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9707855224609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21040, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.1998864397406578, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.01282898411154747, + "grad_norm": 0.08702633529901505, + "kept_token_count": 654.725, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9927360057830811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007263980619609356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.6189331054687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9404998779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21050, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20040369182825088, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.011315437173470855, + "grad_norm": 0.10394575446844101, + "kept_token_count": 654.275, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933792531490326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006620742287486792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.00755615234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.15769653320314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21060, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20026818215847014, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.010192097723484039, + "grad_norm": 0.08458410203456879, + "kept_token_count": 654.421875, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9931847274303436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068152730353176596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.1647247314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.54307250976564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21070, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.2004341796040535, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.014389202278107405, + "grad_norm": 0.09669306874275208, + "kept_token_count": 654.3515625, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9923663139343262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007633682992309332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.92975463867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.38408813476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21080, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.1996421530842781, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.011742047406733036, + "grad_norm": 0.09777867048978806, + "kept_token_count": 654.896875, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9932026743888855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006797332502901554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.39979553222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.960498046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21090, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20049110054969788, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.012328251963481307, + "grad_norm": 0.12624160945415497, + "kept_token_count": 654.2703125, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993217384815216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006782622821629047, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.2237823486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3402801513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21100, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.19973299503326417, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.011846634652465582, + "grad_norm": 0.09083139896392822, + "kept_token_count": 654.7078125, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.993084043264389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006915964651852846, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.7276336669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.858642578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21110, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.20087515115737914, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.01252072243951261, + "grad_norm": 0.09797483682632446, + "kept_token_count": 654.028125, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9925813436508178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007418646477162838, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.3885131835938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.41631774902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21120, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.2001199558377266, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.01345675503835082, + "grad_norm": 0.11829712986946106, + "kept_token_count": 654.6421875, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9930419147014617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006958079151809215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.99997863769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3677154541016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21130, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.20019373893737794, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.011556454794481396, + "grad_norm": 0.08863607794046402, + "kept_token_count": 654.2234375, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9929631650447845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070368263870477675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.4943420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4505187988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21140, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.20084818154573442, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.01194280288182199, + "grad_norm": 0.08625376224517822, + "kept_token_count": 653.7640625, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9932764828205108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006723501905798912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.00087585449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0096466064453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21150, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.1983148619532585, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.010775646893307567, + "grad_norm": 0.09724178165197372, + "kept_token_count": 655.84375, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993075716495514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006924273632466793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5962310791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4383972167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21160, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.2006795287132263, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.012693669088184834, + "grad_norm": 0.10242840647697449, + "kept_token_count": 654.3390625, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9929312705993653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007068741880357265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9268341064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2840057373047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21170, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.20070066303014755, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.012393189687281847, + "grad_norm": 0.10591054707765579, + "kept_token_count": 654.159375, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9929253995418549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007074608001857996, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9042114257812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8765930175781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.20075372010469436, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.013832093821838498, + "grad_norm": 0.1328052282333374, + "kept_token_count": 653.85, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9930760324001312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006923968624323607, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.71848449707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98996887207034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21190, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.1988617479801178, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.011683383630588651, + "grad_norm": 0.07912622392177582, + "kept_token_count": 655.7125, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9928972959518433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071027067489922045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5376281738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4276458740234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21200, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20067958533763885, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.012509614787995815, + "grad_norm": 0.1015133261680603, + "kept_token_count": 653.9234375, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9926605641841888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00733944708481431, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.20061645507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.35028076171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21210, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20023707151412964, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.010679938178509474, + "grad_norm": 0.10753434151411057, + "kept_token_count": 654.4203125, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9928693652153016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007130634784698486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.0006530761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4750274658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20044240206480027, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.01329069104976952, + "grad_norm": 0.11495424062013626, + "kept_token_count": 654.021875, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9929580688476562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007041933014988899, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.9326965332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.875439453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21230, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.20035236924886704, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.012343113543465734, + "grad_norm": 0.12797929346561432, + "kept_token_count": 654.44375, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933825612068177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617448665201664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.35655212402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0810028076172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21240, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.1991522192955017, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.01027267905883491, + "grad_norm": 0.11045771837234497, + "kept_token_count": 655.3015625, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.993124932050705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00687506627291441, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.4821350097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3214508056641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21250, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.19961321949958802, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.0104924778919667, + "grad_norm": 0.10106276720762253, + "kept_token_count": 654.8609375, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9928598642349243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007140146847814321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.54551391601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.08696899414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21260, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.20006705969572067, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.011606335919350386, + "grad_norm": 0.10129521042108536, + "kept_token_count": 654.365625, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933225631713867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006677435059100389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.65105895996095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2117523193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21270, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19930248707532883, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.011376262921839953, + "grad_norm": 0.09874016046524048, + "kept_token_count": 655.09375, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9936025381088257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006397444289177656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.13759155273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.30861206054686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21280, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19998869746923448, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.01161700338125229, + "grad_norm": 0.07835406810045242, + "kept_token_count": 654.753125, + "learning_rate": 6.399965674872228e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9932953238487243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006704685930162668, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.40003662109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3349548339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21290, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2005100131034851, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.012144994782283902, + "grad_norm": 0.10275045782327652, + "kept_token_count": 654.24375, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9931102752685547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889717001467943, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8476928710937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.59253845214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21300, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20081404596567154, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.014145971275866032, + "grad_norm": 0.1310112327337265, + "kept_token_count": 653.971875, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9932556033134461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006744385790079832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.5426483154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9963439941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21310, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.19934160858392716, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.010616892389953137, + "grad_norm": 0.09789294004440308, + "kept_token_count": 655.1703125, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9929175734519958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007082429900765419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.04981689453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0205139160156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21320, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19942618757486344, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.011717168148607016, + "grad_norm": 0.09349354356527328, + "kept_token_count": 655.1609375, + "learning_rate": 6.345860573047314e-06, + "loss": 0.015, + "mgd_cos_sim": 0.993357253074646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642743293195963, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3647064208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3065582275391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21330, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.20016744285821914, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.01202915357425809, + "grad_norm": 0.1130722165107727, + "kept_token_count": 654.215625, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930934250354767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906574312597513, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.34702758789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.3896881103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21340, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.19999546706676483, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.014072018209844828, + "grad_norm": 0.11159129440784454, + "kept_token_count": 654.596875, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0178, + "mgd_cos_sim": 0.9925435245037079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074564795009791855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.02830505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.7196075439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21350, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2007416680455208, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.012538170535117387, + "grad_norm": 0.13654090464115143, + "kept_token_count": 653.859375, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933004200458526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006699590105563402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.15997314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0459442138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21360, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19955333024263383, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.010637315222993493, + "grad_norm": 0.10744280368089676, + "kept_token_count": 654.9015625, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9926897883415222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007310207281261682, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.79706420898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9557739257813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21370, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.1998193383216858, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.01094880849123001, + "grad_norm": 0.11228291690349579, + "kept_token_count": 654.8015625, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9929195284843445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007080455776304007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5204376220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6448120117187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21380, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.20074717700481415, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.01127168396487832, + "grad_norm": 0.10960259288549423, + "kept_token_count": 653.878125, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9927291214466095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007270874828100205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.04859619140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.09344787597655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21390, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2002182424068451, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.01301522394642234, + "grad_norm": 0.11832961440086365, + "kept_token_count": 654.3703125, + "learning_rate": 6.251582125592394e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9919915318489074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008008466195315122, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.19866943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2069885253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21400, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.1994478076696396, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.01224460150115192, + "grad_norm": 0.07977423071861267, + "kept_token_count": 654.9, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9927700161933899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007229986228048801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4313690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7417755126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.2003000095486641, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.012367367604747414, + "grad_norm": 0.09793595224618912, + "kept_token_count": 654.178125, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932021737098694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067978245206177235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.41859130859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9640747070313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21420, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.19850668162107468, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.011957622272893787, + "grad_norm": 0.08897353708744049, + "kept_token_count": 655.8859375, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9929471254348755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00705286543816328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2403503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8240966796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21430, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20067302584648133, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.01235833317041397, + "grad_norm": 0.1197420284152031, + "kept_token_count": 654.0171875, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.993078762292862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069212310016155245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.96007385253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7591247558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21440, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20150080919265748, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.01065390557050705, + "grad_norm": 0.09033181518316269, + "kept_token_count": 653.509375, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9935164391994477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006483559962362051, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8751922607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4734222412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21450, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.20015207678079605, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.012037888215854765, + "grad_norm": 0.11390122771263123, + "kept_token_count": 654.4921875, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9932847380638122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006715274229645729, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.9773284912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.77117614746095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21460, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.20089775919914246, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.01320366570726037, + "grad_norm": 0.09287059307098389, + "kept_token_count": 653.9359375, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931475222110748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006852473597973585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1949798583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.00978698730466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21470, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.1993260458111763, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.012550439033657313, + "grad_norm": 0.0853053405880928, + "kept_token_count": 655.1234375, + "learning_rate": 6.144473460445945e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930211305618286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069788599386811255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.85487670898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.42174072265624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21480, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.20041761696338653, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.012156155984848737, + "grad_norm": 0.08976399153470993, + "kept_token_count": 654.053125, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933119595050812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006688040122389794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6676300048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15936279296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21490, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.20077956318855286, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.01113485749810934, + "grad_norm": 0.07985232025384903, + "kept_token_count": 653.9078125, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933965325355529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006603475380688906, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2234741210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1150726318359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21500, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.19952702820301055, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.012205502158030868, + "grad_norm": 0.12110164016485214, + "kept_token_count": 655.046875, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.993458491563797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006541491113603115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.8198699951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.16819763183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21510, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20074303299188614, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.013422910589724779, + "grad_norm": 0.1215091273188591, + "kept_token_count": 654.0734375, + "learning_rate": 6.091176854918892e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9929188787937164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070811138488352295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.0973175048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.32570495605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21520, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.20017502456903458, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011192511348053814, + "grad_norm": 0.13001622259616852, + "kept_token_count": 654.715625, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9934894144535065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006510574929416179, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.6088165283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7120025634766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21530, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.2000415191054344, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.011701180879026652, + "grad_norm": 0.13888297975063324, + "kept_token_count": 654.6640625, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.992369395494461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007630618568509817, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.81542053222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23284606933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21540, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.2006068781018257, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.012591687450185417, + "grad_norm": 0.10535445064306259, + "kept_token_count": 654.028125, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9930578470230103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006942167412489653, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4652160644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.2218444824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21550, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.20048693269491197, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.012003275984898209, + "grad_norm": 0.0987420454621315, + "kept_token_count": 654.3796875, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9933933973312378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006606616917997599, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2289520263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5605163574219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21560, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.1994297131896019, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.012189353164285422, + "grad_norm": 0.11229025572538376, + "kept_token_count": 655.146875, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9927919745445252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007208019588142633, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.01239624023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.95758666992185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21570, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2008471354842186, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.012384667061269283, + "grad_norm": 0.10049059987068176, + "kept_token_count": 653.96875, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9935553789138794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444627605378628, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.82298889160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.42743530273435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21580, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2000930517911911, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.011382713681086898, + "grad_norm": 0.09070475399494171, + "kept_token_count": 654.6125, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934929192066193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006507066357880831, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.44183654785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15526428222654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21590, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.20010506957769394, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.011661381460726261, + "grad_norm": 0.12608182430267334, + "kept_token_count": 654.6203125, + "learning_rate": 5.98510427876124e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9933132827281952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006686719227582216, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.230126953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3906768798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21600, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.20012417137622834, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.012640720698982477, + "grad_norm": 0.10049968957901001, + "kept_token_count": 654.234375, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9927308738231659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007269134931266308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.869140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1596374511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21610, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.20068830698728563, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.012077206373214721, + "grad_norm": 0.10619495064020157, + "kept_token_count": 654.1421875, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9932766318321228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672336695715785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.75384826660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4173095703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21620, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20103505849838257, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.011692666076123714, + "grad_norm": 0.11329606920480728, + "kept_token_count": 653.9375, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9932403266429901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006759670749306679, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.09784240722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18251037597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21630, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20033162981271743, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.010948208207264543, + "grad_norm": 0.09222598373889923, + "kept_token_count": 654.175, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9930711209774017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006928863842040301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.12115478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2675262451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21640, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.20020408630371095, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.012298460165038706, + "grad_norm": 0.07498592138290405, + "kept_token_count": 654.709375, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931040048599243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00689599709585309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 454.24375915527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.064892578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21650, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19965804368257523, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.010385128809139133, + "grad_norm": 0.0720241442322731, + "kept_token_count": 654.93125, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936821758747101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317822728306055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.8365417480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7388854980469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21660, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.19945951998233796, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.012552146706730128, + "grad_norm": 0.10050418227910995, + "kept_token_count": 654.7046875, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933081150054932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00669186757877469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.67153015136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9904418945313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21670, + "valid_token_count": 817.828125 + }, + { + "actual_token_mask_ratio": 0.200227952003479, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.013312146626412868, + "grad_norm": 0.09174476563930511, + "kept_token_count": 654.5265625, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9929235100746154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007076495327055454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.60578002929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.22027893066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21680, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.199371337890625, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.009949336433783174, + "grad_norm": 0.11694555729627609, + "kept_token_count": 655.28125, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9932950615882874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006704935245215893, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.13307495117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.06740417480466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21690, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.19927479028701783, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.013189203385263682, + "grad_norm": 0.09312981367111206, + "kept_token_count": 655.3125, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931652069091796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006834782660007477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.13544311523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8598205566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21700, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.19967461824417115, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.011604301119223236, + "grad_norm": 0.08428216725587845, + "kept_token_count": 654.8421875, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9926971733570099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007302819099277258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0768768310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.59690856933594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19906091839075088, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.01266358713619411, + "grad_norm": 0.10579884052276611, + "kept_token_count": 655.43125, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931205034255981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006879493780434132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.6965759277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4338745117187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21720, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19998748004436492, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.01198501056060195, + "grad_norm": 0.11596037447452545, + "kept_token_count": 654.346875, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9932288825511932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006771120987832546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.39111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.07830810546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21730, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.20045648962259294, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.010521518625319004, + "grad_norm": 0.07478582859039307, + "kept_token_count": 654.23125, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9932988166809082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006701168324798346, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0873687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.43810119628904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21740, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2000728577375412, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.013813681667670608, + "grad_norm": 0.16046175360679626, + "kept_token_count": 654.3421875, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0171, + "mgd_cos_sim": 0.993364405632019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006635599490255118, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.8663665771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5079803466797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21750, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.2003379449248314, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.011832221224904061, + "grad_norm": 0.0866367518901825, + "kept_token_count": 654.446875, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9929385840892792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007061425689607858, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 453.94016723632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0605102539063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21760, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.20056163668632507, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.01157408207654953, + "grad_norm": 0.12985143065452576, + "kept_token_count": 654.15625, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933775067329407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006622498948127031, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4750610351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.43621826171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21770, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19948381185531616, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.01322530498728156, + "grad_norm": 0.11278131604194641, + "kept_token_count": 655.0078125, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.99310861825943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006891386117786169, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.45737609863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4039794921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21780, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.20051145553588867, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.00954871205613017, + "grad_norm": 0.07418868690729141, + "kept_token_count": 654.2359375, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9935522198677063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064477682113647464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.59454040527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21889038085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.19988062083721161, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.012127484101802111, + "grad_norm": 0.10852976143360138, + "kept_token_count": 654.621875, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9929700493812561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007029965240508318, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.47637634277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8828857421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21800, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.19937180429697038, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.013190741976723075, + "grad_norm": 0.11355926841497421, + "kept_token_count": 655.1875, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931214570999145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068785474635660645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.6976776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1813629150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1997173845767975, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.012209058227017521, + "grad_norm": 0.09909415990114212, + "kept_token_count": 654.9875, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9929606914520264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007039305847138166, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.3455505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68236083984374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21820, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20005575567483902, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.012806202983483672, + "grad_norm": 0.09099870175123215, + "kept_token_count": 654.3734375, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933978199958802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006602196581661701, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.35768127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2240753173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21830, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.20016914159059523, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.011431959550827741, + "grad_norm": 0.11526989936828613, + "kept_token_count": 654.3015625, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9932377278804779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006762286834418774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.5234405517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7871063232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21840, + "valid_token_count": 818.05625 + }, + { + "actual_token_mask_ratio": 0.20005830228328705, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.013145213294774295, + "grad_norm": 0.08680961281061172, + "kept_token_count": 654.7140625, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9934493660926819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006550620961934328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.7281829833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3557830810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21850, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.19986060559749602, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.012787852203473449, + "grad_norm": 0.10522235184907913, + "kept_token_count": 654.6296875, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9930797517299652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069202413782477375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.3650634765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2270751953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21860, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.19987686425447465, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.012731043621897697, + "grad_norm": 0.09534170478582382, + "kept_token_count": 654.8375, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9932220995426178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006777889654040336, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.5617340087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9750518798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21870, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19997516572475432, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.012596350256353616, + "grad_norm": 0.11410108953714371, + "kept_token_count": 654.6390625, + "learning_rate": 5.61942465599249e-06, + "loss": 0.016, + "mgd_cos_sim": 0.993243545293808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006756464391946793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.94965209960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3263000488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1994010955095291, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.012578514078631997, + "grad_norm": 0.13228440284729004, + "kept_token_count": 655.371875, + "learning_rate": 5.60652788608023e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9931188344955444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006881173886358738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.12763061523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.53705444335935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21890, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.20050515979528427, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.011192472884431481, + "grad_norm": 0.05976714566349983, + "kept_token_count": 654.2703125, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9930799126625061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006920079793781042, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.40582275390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.19948120117186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21900, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20008096545934678, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.011945120338350535, + "grad_norm": 0.11944539844989777, + "kept_token_count": 654.5484375, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934071481227875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006592855509370566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1517791748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4348876953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21910, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19992404580116271, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.01125743710435927, + "grad_norm": 0.1274464875459671, + "kept_token_count": 654.66875, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9932123899459839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006787599064409733, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.13450317382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57435302734376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19987935423851014, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.014288907404989004, + "grad_norm": 0.10331649333238602, + "kept_token_count": 654.8140625, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9931210875511169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006878911517560482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.69683227539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.304541015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21930, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.20016933232545853, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.01314902352169156, + "grad_norm": 0.10570251196622849, + "kept_token_count": 654.55, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9927797555923462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007220242358744145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.49859008789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1884796142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21940, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2004229873418808, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.012291760928928853, + "grad_norm": 0.1118331104516983, + "kept_token_count": 654.4890625, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9929708361625671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00702917343005538, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 455.3906188964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.12159118652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21950, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19983142912387847, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.011617467552423478, + "grad_norm": 0.1027921736240387, + "kept_token_count": 654.8015625, + "learning_rate": 5.516570963231058e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9932929217815399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006707063503563404, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.6438903808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.12546691894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21960, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19979038536548616, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.011926677823066712, + "grad_norm": 0.08207831531763077, + "kept_token_count": 654.86875, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934474647045135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552543956786394, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3307220458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9785858154297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21970, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19940610975027084, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.013189213909208774, + "grad_norm": 0.08480767905712128, + "kept_token_count": 655.0703125, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9926248967647553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007375103794038296, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3219787597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1315490722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2000194236636162, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.011750978976488113, + "grad_norm": 0.09428869187831879, + "kept_token_count": 654.3671875, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931799054145813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006820099335163832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.8427490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.57440490722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 21990, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19968871921300888, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.012566386535763741, + "grad_norm": 0.11484652757644653, + "kept_token_count": 654.803125, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9930951297283173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006904866825789213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.80882263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3231262207031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22000, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20034738928079604, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.011842711782082915, + "grad_norm": 0.13288560509681702, + "kept_token_count": 654.340625, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9927548587322235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072451353073120115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7357666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.01373291015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22010, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20056209564208985, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.012327913613989949, + "grad_norm": 0.13068482279777527, + "kept_token_count": 654.1640625, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9934621095657349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537895929068327, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.7626708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2565032958984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22020, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20055943727493286, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.011588582210242749, + "grad_norm": 0.11933942884206772, + "kept_token_count": 653.9921875, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9934794902801514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520505994558334, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.5253112792969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.23663940429685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22030, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.1990411475300789, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.01195009909570217, + "grad_norm": 0.1568925976753235, + "kept_token_count": 655.3796875, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9930333614349365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006966633908450604, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.5501647949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.88555603027345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20010401010513307, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.012249027099460363, + "grad_norm": 0.08662468940019608, + "kept_token_count": 654.5078125, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9936011850833892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006398812122642994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.7692443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.614501953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22050, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.1998821884393692, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.00912521332502365, + "grad_norm": 0.10224977880716324, + "kept_token_count": 654.55625, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.993379020690918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006620971392840147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.55579833984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.56395568847654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22060, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19982323050498962, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.011837857263162731, + "grad_norm": 0.08801411092281342, + "kept_token_count": 654.5625, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9930988907814026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006901117227971554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.5775543212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3265655517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22070, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.1991572931408882, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.011977990251034498, + "grad_norm": 0.08719048649072647, + "kept_token_count": 655.4328125, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993194854259491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006805149093270302, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1482727050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0997039794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22080, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20064881443977356, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.01065481654368341, + "grad_norm": 0.12447823584079742, + "kept_token_count": 654.0859375, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9935382664203644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006461727526038885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2008026123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9765960693359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22090, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20026711225509644, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.011573633039370179, + "grad_norm": 0.07217350602149963, + "kept_token_count": 654.5921875, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9930047750473022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006995221320539713, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2095458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0869567871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22100, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20018121749162673, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.013072403194382787, + "grad_norm": 0.11877698451280594, + "kept_token_count": 654.7234375, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9927776515483856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007222344633191824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7639526367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.87498168945314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22110, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.20017287880182266, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.009969537099823355, + "grad_norm": 0.09843665361404419, + "kept_token_count": 654.4828125, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933107852935791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006689216289669275, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.85541076660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9336273193359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22120, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.20015800148248672, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.013200408359989525, + "grad_norm": 0.0904141217470169, + "kept_token_count": 654.875, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9931293249130249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068706943653523925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.2617706298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5415100097656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22130, + "valid_token_count": 818.75625 + }, + { + "actual_token_mask_ratio": 0.19861435443162917, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.011577481450513006, + "grad_norm": 0.09537225216627121, + "kept_token_count": 655.6484375, + "learning_rate": 5.287858672567979e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9932228207588196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006777177285403013, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0042358398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.61497192382814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22140, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.20043099224567412, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.011260640621185303, + "grad_norm": 0.09478671103715897, + "kept_token_count": 654.05625, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.993155550956726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006844462547451258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.29005737304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1772735595703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22150, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.20043869465589523, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.014735636953264474, + "grad_norm": 0.07126136869192123, + "kept_token_count": 654.39375, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0182, + "mgd_cos_sim": 0.9930771231651306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922870967537164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.4898956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9719543457031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22160, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.20080227106809617, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.012048402847722173, + "grad_norm": 0.11311444640159607, + "kept_token_count": 653.9703125, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936229526996613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006377037614583969, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.64369812011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3667694091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22170, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20027669817209243, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.011528262635692954, + "grad_norm": 0.08666551113128662, + "kept_token_count": 654.2140625, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9933704555034637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006629544869065284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8168518066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9093475341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22180, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20029767006635665, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.010597193799912929, + "grad_norm": 0.07482558488845825, + "kept_token_count": 654.20625, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9928371906280518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00716281458735466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5630615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.32842712402345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22190, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.19990879893302918, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.012109092017635704, + "grad_norm": 0.11557304114103317, + "kept_token_count": 654.6296875, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9932273983955383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006772597692906857, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.07989807128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94273071289064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22200, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.1992976799607277, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.014284901320934296, + "grad_norm": 0.10407311469316483, + "kept_token_count": 655.2203125, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0179, + "mgd_cos_sim": 0.9927871882915497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007212808635085821, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.6944091796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3404510498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22210, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.20009772777557372, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.01030808873474598, + "grad_norm": 0.09124328941106796, + "kept_token_count": 654.5390625, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936105310916901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006389477755874395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.87265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.78550109863284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22220, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19956808239221574, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.011019758880138397, + "grad_norm": 0.0983923003077507, + "kept_token_count": 654.79375, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9937475979328155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006252405699342489, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5240478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7314727783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.20061210095882415, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.0106887626927346, + "grad_norm": 0.1026536375284195, + "kept_token_count": 654.234375, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9930582761764526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941720563918352, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.35008544921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0769409179687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22240, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19979291558265685, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.011036523943766951, + "grad_norm": 0.08932830393314362, + "kept_token_count": 654.6484375, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993749338388443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006250650342553854, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2663269042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22250, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2005743369460106, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.010799268819391727, + "grad_norm": 0.13179758191108704, + "kept_token_count": 654.2546875, + "learning_rate": 5.137502456402705e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9935156524181366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006484362110495567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.09398498535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.52691345214845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22260, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.19943253844976425, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.012402038741856813, + "grad_norm": 0.1020815372467041, + "kept_token_count": 654.9734375, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931015133857727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898488197475672, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.75425415039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0556671142578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22270, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20034119784832, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.01308066314086318, + "grad_norm": 0.1269279420375824, + "kept_token_count": 654.0875, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.992606258392334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007393740769475699, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.640771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.06398620605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22280, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.19901612102985383, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.01056992202065885, + "grad_norm": 0.08448271453380585, + "kept_token_count": 655.4890625, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9930044949054718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006995508447289467, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9345275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.21791381835936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22290, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19946671575307845, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.013396086543798447, + "grad_norm": 0.07816796004772186, + "kept_token_count": 654.946875, + "learning_rate": 5.087765905286004e-06, + "loss": 0.017, + "mgd_cos_sim": 0.992806589603424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071934032253921035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9424621582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.484619140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22300, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.1998215138912201, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.01187801151536405, + "grad_norm": 0.08497379720211029, + "kept_token_count": 654.8125, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9930407047271729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00695930402725935, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2290924072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3824462890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22310, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1996320143342018, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.011753239156678318, + "grad_norm": 0.1126784160733223, + "kept_token_count": 654.84375, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9929474115371704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007052586879581213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5395965576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.65057678222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22320, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20065896064043046, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013433996168896557, + "grad_norm": 0.07214922457933426, + "kept_token_count": 654.2828125, + "learning_rate": 5.05058990415731e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9928193628787995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007180623430758715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.93724670410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.76292114257814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22330, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.19981476068496704, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.012176222819834947, + "grad_norm": 0.08214807510375977, + "kept_token_count": 654.434375, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933600664138794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639923527836799, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.51964111328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3586669921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22340, + "valid_token_count": 817.8546875 + }, + { + "actual_token_mask_ratio": 0.20019487887620926, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.013311382802203297, + "grad_norm": 0.12874265015125275, + "kept_token_count": 654.584375, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9932435631752015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006756454799324274, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3857940673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.30083618164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22350, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.20058038234710693, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.010263867024332286, + "grad_norm": 0.08233815431594849, + "kept_token_count": 654.028125, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9933094263076783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066905728541314605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.1930206298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.63054809570315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22360, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2002829059958458, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.012850621435791253, + "grad_norm": 0.10131068527698517, + "kept_token_count": 654.2515625, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9931413114070893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006858678441494704, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2177703857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4084747314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22370, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.19986069351434707, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.011429676320403814, + "grad_norm": 0.10686817765235901, + "kept_token_count": 654.4953125, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9926343619823456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073656546883285046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5019561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9152893066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22380, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.1996508464217186, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.0130331224296242, + "grad_norm": 0.09945115447044373, + "kept_token_count": 655.2515625, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9937330424785614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006266943737864494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.47197875976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.04282531738284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22390, + "valid_token_count": 818.7109375 + }, + { + "actual_token_mask_ratio": 0.19947097450494766, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.012309692148119212, + "grad_norm": 0.10486035794019699, + "kept_token_count": 655.290625, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.992762702703476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072372954338788984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4900421142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.27369689941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22400, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.20066188275814056, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.013120264001190662, + "grad_norm": 0.09589390456676483, + "kept_token_count": 653.7609375, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9933260321617127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00667396979406476, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.08013610839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.62890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22410, + "valid_token_count": 817.8734375 + }, + { + "actual_token_mask_ratio": 0.2002677544951439, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.013164743781089783, + "grad_norm": 0.0823177695274353, + "kept_token_count": 654.7296875, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9932527363300323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747281365096569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.65155029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1984893798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22420, + "valid_token_count": 818.69375 + }, + { + "actual_token_mask_ratio": 0.19975328892469407, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.011471191421151161, + "grad_norm": 0.10310263186693192, + "kept_token_count": 654.9953125, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9931216955184936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006878304574638605, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.05080261230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30784912109374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22430, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.20010311752557755, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.012507519498467446, + "grad_norm": 0.11671806871891022, + "kept_token_count": 654.4703125, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934469580650329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006553047336637974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.96068115234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6628143310547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22440, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.2000093564391136, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.013353596394881605, + "grad_norm": 0.06879104673862457, + "kept_token_count": 654.5359375, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933116674423218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006688348669558764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.2571136474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2915374755859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22450, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2001645654439926, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.013136998005211353, + "grad_norm": 0.0677385926246643, + "kept_token_count": 654.5015625, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9937140107154846, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006285984721034765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.4233093261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.74444274902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22460, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1987040936946869, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010056615713983775, + "grad_norm": 0.09891609102487564, + "kept_token_count": 655.55625, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9931990623474121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006800942961126566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0371398925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3353210449219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22470, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.19979100823402404, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.010414775880053639, + "grad_norm": 0.08898518979549408, + "kept_token_count": 654.78125, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9932040750980378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067959165200591086, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2367706298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0169250488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22480, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20012269020080567, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.010009940713644028, + "grad_norm": 0.09755335003137589, + "kept_token_count": 654.5578125, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933447480201721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006655251700431108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.39418029785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2207763671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22490, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.2001072958111763, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.011370921926572918, + "grad_norm": 0.11071611940860748, + "kept_token_count": 654.496875, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.993373042345047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00662694638594985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1611083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.25205383300784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22500, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.20041577517986298, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.012326620053499937, + "grad_norm": 0.10897110402584076, + "kept_token_count": 654.128125, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9930444777011871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006955529563128948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.82924499511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5266967773438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22510, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2002690017223358, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013386929221451283, + "grad_norm": 0.130051851272583, + "kept_token_count": 654.4015625, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9930771291255951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922886613756419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.3291046142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8874145507813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22520, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.1996827483177185, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.009775870991870762, + "grad_norm": 0.09214845299720764, + "kept_token_count": 654.875, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9931600213050842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006839981209486723, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1542663574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.73247680664065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22530, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19946246296167375, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.012587806908413768, + "grad_norm": 0.1105843111872673, + "kept_token_count": 654.696875, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933806955814362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006619316432625056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5398895263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1288818359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22540, + "valid_token_count": 817.81875 + }, + { + "actual_token_mask_ratio": 0.19958345592021942, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.01191742871887982, + "grad_norm": 0.0811077430844307, + "kept_token_count": 654.60625, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9929686665534974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031335402280093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1483978271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7230590820312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22550, + "valid_token_count": 817.8328125 + }, + { + "actual_token_mask_ratio": 0.19996441453695296, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.010392655245959758, + "grad_norm": 0.08329488337039948, + "kept_token_count": 654.6171875, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9933863341808319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066136734560132025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0526153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0876525878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22560, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20106441974639894, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.0113521633669734, + "grad_norm": 0.09534277766942978, + "kept_token_count": 653.6609375, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.993013721704483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006986274663358927, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.42164611816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15679321289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19943429827690123, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.012588579673320055, + "grad_norm": 0.12388968467712402, + "kept_token_count": 655.2671875, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9924839496612549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075160423293709755, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6232421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0654022216797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22580, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.19925976395606995, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.01155296815559268, + "grad_norm": 0.1129051148891449, + "kept_token_count": 655.0734375, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.993207722902298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006792272627353668, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4591949462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.533447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22590, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.19962482303380966, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.010608042916283011, + "grad_norm": 0.09691105782985687, + "kept_token_count": 654.8609375, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.993354332447052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006645669788122177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3214172363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1641326904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22600, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19960968494415282, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.01328761586919427, + "grad_norm": 0.08991487324237823, + "kept_token_count": 655.15625, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9926042258739471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007395768444985152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.36463317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.98427429199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22610, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.1993392899632454, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.012956443149596453, + "grad_norm": 0.1033661738038063, + "kept_token_count": 655.028125, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9927768647670746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007223141938447952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.34808349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8083190917969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22620, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2003837689757347, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.012545470613986254, + "grad_norm": 0.08101505786180496, + "kept_token_count": 654.4890625, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9924915254116058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007508465368300676, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.9652160644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3183197021484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22630, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.20001475363969803, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.010636051185429096, + "grad_norm": 0.09768474847078323, + "kept_token_count": 654.4609375, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9927200853824616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007279906049370765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3112365722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4431884765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22640, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.20078288912773132, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.012790935952216387, + "grad_norm": 0.08504655212163925, + "kept_token_count": 653.6234375, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9930087924003601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069912165403366085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6543334960937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.62115478515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22650, + "valid_token_count": 817.83125 + }, + { + "actual_token_mask_ratio": 0.20023522824048995, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.012667804071679712, + "grad_norm": 0.11971765011548996, + "kept_token_count": 654.484375, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932881534099579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006711851619184017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.87486267089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9637725830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22660, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.2009490966796875, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.013007963309064507, + "grad_norm": 0.08393923193216324, + "kept_token_count": 653.8984375, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9925200521945954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007479952834546566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9438079833984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6954803466797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22670, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.1999339461326599, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.01150494241155684, + "grad_norm": 0.10475374013185501, + "kept_token_count": 654.70625, + "learning_rate": 4.625015114690474e-06, + "loss": 0.015, + "mgd_cos_sim": 0.992937046289444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007062955480068922, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.82236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.54344177246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22680, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20069287568330765, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.013832229422405362, + "grad_norm": 0.09500589221715927, + "kept_token_count": 654.1171875, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.993012422323227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006987582053989172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.2619659423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.90123291015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22690, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.20040389001369477, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.01216585747897625, + "grad_norm": 0.10092134773731232, + "kept_token_count": 654.1265625, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9930131077766419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00698689017444849, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1230743408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.81830444335935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22700, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.1994515150785446, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.013208122923970222, + "grad_norm": 0.1059039980173111, + "kept_token_count": 654.9046875, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9929644703865051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007035544142127037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4759796142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5356689453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22710, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.20098810344934465, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.013093943521380425, + "grad_norm": 0.11048483848571777, + "kept_token_count": 653.9578125, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9926665902137757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007333398796617985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.629541015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7083282470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22720, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.2002387598156929, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.010360206197947263, + "grad_norm": 0.11360824853181839, + "kept_token_count": 654.421875, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9928375780582428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007162441313266754, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.49615173339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8721221923828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22730, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.19999346882104874, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.013439785921946168, + "grad_norm": 0.0894768163561821, + "kept_token_count": 654.3640625, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9937301874160767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062698150984942915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.3194183349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3967010498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22740, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.1995578721165657, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.012399771576747298, + "grad_norm": 0.11086926609277725, + "kept_token_count": 655.065625, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9926320850849152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073679198510944845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.9717498779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8048156738281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22750, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.1998743385076523, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012394016701728105, + "grad_norm": 0.08603866398334503, + "kept_token_count": 654.60625, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933442354202271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006655767653137445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.3956268310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7902404785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22760, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.20043076872825621, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.011881728703156114, + "grad_norm": 0.0843811109662056, + "kept_token_count": 654.196875, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9932406902313232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006759307812899351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.69313659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3213745117188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22770, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.1996914342045784, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.012563478900119662, + "grad_norm": 0.11073268949985504, + "kept_token_count": 654.8875, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9927045702934265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007295422255992889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.6154327392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.08978576660155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20052033811807632, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.011760914977639914, + "grad_norm": 0.07735607028007507, + "kept_token_count": 653.99375, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9937747061252594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225292664021253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.70452880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0093017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22790, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20065760463476182, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.009718256816267968, + "grad_norm": 0.08293682336807251, + "kept_token_count": 654.2796875, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9935833752155304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006416634004563093, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.0476043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5579132080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22800, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.20010706633329392, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.011229654122143985, + "grad_norm": 0.11267013847827911, + "kept_token_count": 654.4515625, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933348894119263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00666512493044138, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.4953369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8161651611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22810, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.20063498467206956, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.010082605620846152, + "grad_norm": 0.06517406553030014, + "kept_token_count": 654.1296875, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9930027186870575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997284945100546, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4338714599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7771484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22820, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.19979197680950164, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.009476609574630857, + "grad_norm": 0.06230965256690979, + "kept_token_count": 655.1265625, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9935561358928681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443866062909365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.77842407226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.99620056152344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22830, + "valid_token_count": 818.6921875 + }, + { + "actual_token_mask_ratio": 0.2000299498438835, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.012046121852472424, + "grad_norm": 0.13571499288082123, + "kept_token_count": 654.3640625, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.992728316783905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007271671108901501, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0583190917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.23414916992186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22840, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.20028141140937805, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.011182133993133902, + "grad_norm": 0.11818134039640427, + "kept_token_count": 654.4546875, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9928392767906189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071607263758778576, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0828887939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.49876098632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.20048698484897615, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.01134222922846675, + "grad_norm": 0.059082187712192535, + "kept_token_count": 654.365625, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9931516766548156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006848311983048916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9806427001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8889892578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22860, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.20020158141851424, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.013695189822465181, + "grad_norm": 0.1174970269203186, + "kept_token_count": 654.128125, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0174, + "mgd_cos_sim": 0.9925283908843994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007471607904881239, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.52984313964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8895965576172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22870, + "valid_token_count": 817.8671875 + }, + { + "actual_token_mask_ratio": 0.2006734788417816, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.012814374221488834, + "grad_norm": 0.07447195798158646, + "kept_token_count": 654.0046875, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9934003412723541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006599656585603952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.37001647949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15293884277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22880, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20001738965511323, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.012387655954807997, + "grad_norm": 0.12410122901201248, + "kept_token_count": 654.371875, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932197034358978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006780287344008684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5072998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5316650390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22890, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.20010267645120622, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.011268694698810578, + "grad_norm": 0.05499636009335518, + "kept_token_count": 654.509375, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934000372886658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006599960476160049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9523986816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.124609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22900, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.20040446072816848, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.01152390157803893, + "grad_norm": 0.08441077917814255, + "kept_token_count": 654.246875, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9935015499591827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006498468760401011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9440948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0345947265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.2002757892012596, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.012285011447966098, + "grad_norm": 0.08889292925596237, + "kept_token_count": 654.203125, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9933121979236603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006687804218381643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.9686340332031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.85817565917966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22920, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.1993489533662796, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.012801958713680506, + "grad_norm": 0.07201936841011047, + "kept_token_count": 655.1890625, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933215141296386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006678491830825806, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1074554443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5885894775391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22930, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.19999155402183533, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.011885984288528561, + "grad_norm": 0.14291340112686157, + "kept_token_count": 654.5921875, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9929980993270874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070019039325416085, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.3533599853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6850646972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22940, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.1995447963476181, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.010328260576352477, + "grad_norm": 0.09521663188934326, + "kept_token_count": 654.9921875, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.993555074930191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444917060434818, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.1863128662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8632843017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2005183979868889, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.010192720498889685, + "grad_norm": 0.09063815325498581, + "kept_token_count": 654.1109375, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9933938145637512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006606200430542231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.02703247070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.474365234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22960, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20027133226394653, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.011730693792924286, + "grad_norm": 0.1328320950269699, + "kept_token_count": 654.465625, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9929273545742034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007072644215077162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.07505798339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.62976989746096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22970, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2006359040737152, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.012050039460882544, + "grad_norm": 0.08811215311288834, + "kept_token_count": 654.6875, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930478930473328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006952080596238375, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.19517517089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84198303222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22980, + "valid_token_count": 819.003125 + }, + { + "actual_token_mask_ratio": 0.20057859569787978, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.013240993116050958, + "grad_norm": 0.15614956617355347, + "kept_token_count": 654.2515625, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9930818438529968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006918157637119293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.91014404296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.49073486328126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 22990, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.20066132843494416, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.011208440642803907, + "grad_norm": 0.08548370003700256, + "kept_token_count": 654.01875, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9932984232902526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00670157577842474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.56849060058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6903900146484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23000, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.19949132800102234, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.01137258424423635, + "grad_norm": 0.12980128824710846, + "kept_token_count": 654.925, + "learning_rate": 4.237890962316264e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9927184879779816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00728151211515069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.44884033203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6477325439453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23010, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.20132029205560684, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.011350061092525721, + "grad_norm": 0.08669386804103851, + "kept_token_count": 653.48125, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9930932283401489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906767655164003, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.7874786376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.43626098632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23020, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20025087743997574, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.012675736658275127, + "grad_norm": 0.09560537338256836, + "kept_token_count": 654.3, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931037247180938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006896270625293255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6727264404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.24598693847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23030, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.1994022935628891, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.012474110489711166, + "grad_norm": 0.09370272606611252, + "kept_token_count": 655.128125, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9934275448322296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006572465971112251, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3468444824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2152618408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23040, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20057820975780488, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.014037310378625989, + "grad_norm": 0.12307732552289963, + "kept_token_count": 654.1390625, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0176, + "mgd_cos_sim": 0.992926961183548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007073037791997194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2943145751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5931701660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23050, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.19936142414808272, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.011397260101512074, + "grad_norm": 0.10338761657476425, + "kept_token_count": 655.2859375, + "learning_rate": 4.180465573027335e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9927939057350159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007206092122942209, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4417785644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6723297119141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23060, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.19969182461500168, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.011304592154920102, + "grad_norm": 0.09813077747821808, + "kept_token_count": 655.1125, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9927572309970856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007242768164724111, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9399658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8735107421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23070, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.20145806074142455, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.012076869793236255, + "grad_norm": 0.09568341076374054, + "kept_token_count": 653.540625, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9927648842334748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007235124427825213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6484313964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2831665039063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23080, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.20038863718509675, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.011482869600877165, + "grad_norm": 0.09159381687641144, + "kept_token_count": 654.2859375, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9930774688720703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922523025423288, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.72747192382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7508972167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23090, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.2004635751247406, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.01245995149947703, + "grad_norm": 0.09625577181577682, + "kept_token_count": 654.353125, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932979106903076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006702099274843931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0415466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2858001708984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23100, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19927651435136795, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.0128424477763474, + "grad_norm": 0.09434002637863159, + "kept_token_count": 655.371875, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9933436512947083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006656346470117569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1831359863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2157775878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23110, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.19955629706382752, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.011113724764436483, + "grad_norm": 0.08492540568113327, + "kept_token_count": 654.9890625, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933091819286346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006690829712897539, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.55091857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0326690673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23120, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.19911666661500932, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.010839385725557805, + "grad_norm": 0.0925304964184761, + "kept_token_count": 655.453125, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934675216674804, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006532467436045408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9511413574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.76047668457034, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23130, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.1990965873003006, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.010292638931423426, + "grad_norm": 0.0930585116147995, + "kept_token_count": 655.546875, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9929754197597503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007024580147117376, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4750152587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.37293701171876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23140, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.20116790384054184, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.013345594890415669, + "grad_norm": 0.1191154271364212, + "kept_token_count": 653.8171875, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9933244824409485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006675503682345152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.65870361328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.933740234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.19919995665550233, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.013666291069239379, + "grad_norm": 0.11760061234235764, + "kept_token_count": 655.30625, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0173, + "mgd_cos_sim": 0.9927944660186767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072055417113006115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.2462982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0747589111328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23160, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.1995905965566635, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.010449987789615988, + "grad_norm": 0.08193761855363846, + "kept_token_count": 654.61875, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9931756019592285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006824397761374712, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.802490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.02520751953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23170, + "valid_token_count": 817.85625 + }, + { + "actual_token_mask_ratio": 0.19931321889162062, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.011741699650883675, + "grad_norm": 0.06458448618650436, + "kept_token_count": 655.3125, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933335840702057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006666419375687837, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.12415771484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.39202575683595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23180, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19941516369581222, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.012676917016506195, + "grad_norm": 0.10059227049350739, + "kept_token_count": 655.3828125, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9931291162967681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006870878674089908, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6077575683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3957061767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23190, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.20114711821079254, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.010741621116176247, + "grad_norm": 0.11307157576084137, + "kept_token_count": 653.8328125, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9931388080120087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006861190125346184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0039306640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.14064025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.19942685216665268, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.010226963041350245, + "grad_norm": 0.10312587022781372, + "kept_token_count": 655.353125, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9934827446937561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006517268437892199, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.19434814453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.91431579589846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23210, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.20091697722673416, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.012362133804708719, + "grad_norm": 0.09226090461015701, + "kept_token_count": 653.8703125, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9933759152889252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006624081265181303, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0412170410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3313232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23220, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.19920527338981628, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.011079828068614007, + "grad_norm": 0.09608728438615799, + "kept_token_count": 655.1125, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934669375419617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533067207783461, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.65020141601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1711669921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23230, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.20025441646575928, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.012422990752384067, + "grad_norm": 0.10122810304164886, + "kept_token_count": 654.5703125, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932370245456695, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006762972380965948, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.054248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1805816650391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23240, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.20033127218484878, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.011425998294726014, + "grad_norm": 0.07547789067029953, + "kept_token_count": 654.275, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9923290908336639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076708959415555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.4401489257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.57973327636716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23250, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20000526160001755, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.014143685391172767, + "grad_norm": 0.07522492855787277, + "kept_token_count": 654.5140625, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0177, + "mgd_cos_sim": 0.9928205132484436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007179478276520967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.02796325683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.74790344238284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23260, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2000262290239334, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.010332795837894083, + "grad_norm": 0.07717350870370865, + "kept_token_count": 654.5765625, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9929815351963043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007018476910889149, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1588836669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.03639221191406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23270, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20066273510456084, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.012525849789381028, + "grad_norm": 0.08187729120254517, + "kept_token_count": 653.98125, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9927150428295135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007284949161112308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.89171447753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.810791015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.20037871748209, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.010529844556003809, + "grad_norm": 0.08526850491762161, + "kept_token_count": 654.5078125, + "learning_rate": 3.920568733528047e-06, + "loss": 0.014, + "mgd_cos_sim": 0.993039608001709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006960389204323292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7030487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84886474609374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23290, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.20054855048656464, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.011275645019486547, + "grad_norm": 0.10151843726634979, + "kept_token_count": 654.196875, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936067700386048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006393230240792036, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3383026123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9486572265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23300, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.20017519295215608, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.011870058113709092, + "grad_norm": 0.07618145644664764, + "kept_token_count": 654.5515625, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9934534430503845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006546555738896132, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.6184478759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8766754150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23310, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.20100072622299195, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.009590957080945373, + "grad_norm": 0.0891350656747818, + "kept_token_count": 653.9375, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9934790551662445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520947720855474, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.2471405029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3668701171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23320, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.20009148120880127, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.010682687116786837, + "grad_norm": 0.0712367445230484, + "kept_token_count": 654.5265625, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9931864321231842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006813568435609341, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.6985198974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.32354736328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23330, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19950905442237854, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.013435101741924882, + "grad_norm": 0.10779841244220734, + "kept_token_count": 654.93125, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9930769264698028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006923081912100315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.65479736328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.33525390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23340, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2001582518219948, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.01090596904978156, + "grad_norm": 0.1229003295302391, + "kept_token_count": 654.4421875, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9932656764984131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00673433281481266, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7925109863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3938690185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19876337200403213, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.011545421555638313, + "grad_norm": 0.08370889723300934, + "kept_token_count": 655.4, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9930025219917298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997475307434798, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.26670227050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.63970947265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23360, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.1996096268296242, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.012861057743430137, + "grad_norm": 0.12079906463623047, + "kept_token_count": 654.9, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.992610615491867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007389376033097505, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1535217285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.55995788574216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23370, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.1997319221496582, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.011800199141725897, + "grad_norm": 0.07338320463895798, + "kept_token_count": 654.909375, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9937477290630341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006252262368798256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.18781127929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5426544189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23380, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.20024365484714507, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.011463673133403064, + "grad_norm": 0.09575805068016052, + "kept_token_count": 654.146875, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9927838027477265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00721619613468647, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.98598022460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.10210571289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23390, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.19935886561870575, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.010996145335957408, + "grad_norm": 0.11468444019556046, + "kept_token_count": 655.1609375, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9936822533607483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063177491538226604, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.777880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4302093505859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23400, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19952357113361358, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.011448835069313646, + "grad_norm": 0.06811532378196716, + "kept_token_count": 654.984375, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9933286011219025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006671409681439399, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7207794189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15235290527346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23410, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.1991847574710846, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.012701936718076467, + "grad_norm": 0.08115938305854797, + "kept_token_count": 655.4953125, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9935843884944916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006415603216737509, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.86986389160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8155578613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23420, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.19950316697359086, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.012171477219089866, + "grad_norm": 0.06507217139005661, + "kept_token_count": 654.8484375, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9931593656539917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006840633973479271, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9950866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.275634765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23430, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.19962544590234757, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.011295013595372438, + "grad_norm": 0.08996915072202682, + "kept_token_count": 654.6578125, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935219049453735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006478095520287752, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5116241455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.41302185058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23440, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.2005395770072937, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.0106951127294451, + "grad_norm": 0.11521921306848526, + "kept_token_count": 654.3, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9929634273052216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007036573812365532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.56390380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4932098388672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23450, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.1996619865298271, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.013615694921463728, + "grad_norm": 0.08815465867519379, + "kept_token_count": 654.76875, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9932938158512116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006706168968230486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.31904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7599212646484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23460, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.19980575144290924, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.0132290942594409, + "grad_norm": 0.07324868440628052, + "kept_token_count": 654.621875, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9935767471790313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006423246208578348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.70237121582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.87001037597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23470, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.2008558616042137, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.012816955242305994, + "grad_norm": 0.09937555342912674, + "kept_token_count": 653.9359375, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9929973602294921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007002629805356264, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.58257446289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.48732604980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23480, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.1998469665646553, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.010997997783124446, + "grad_norm": 0.07443352043628693, + "kept_token_count": 654.734375, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9930009365081787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006999056227505207, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.73504943847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.56605529785156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23490, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19957662969827653, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.012377577275037766, + "grad_norm": 0.08340473473072052, + "kept_token_count": 654.8390625, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9935356616973877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006464337650686503, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.80345458984374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.67820739746094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23500, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.20011079907417298, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.011048394115641713, + "grad_norm": 0.08331397920846939, + "kept_token_count": 654.46875, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9937271356582642, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006272852979600429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.9784393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7606506347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23510, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.2001700669527054, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.01272974517196417, + "grad_norm": 0.11070220172405243, + "kept_token_count": 654.628125, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937409818172455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00625900886952877, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.26304016113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.78546752929685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23520, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.19972058832645417, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.012936405465006828, + "grad_norm": 0.07906968891620636, + "kept_token_count": 655.034375, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932547330856323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006745276972651482, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.9192352294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.19720458984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23530, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.20003570318222047, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.012857276108115911, + "grad_norm": 0.10344217717647552, + "kept_token_count": 654.978125, + "learning_rate": 3.646154645454948e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9937077403068543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062922507524490355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2185424804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7159057617188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23540, + "valid_token_count": 818.759375 + }, + { + "actual_token_mask_ratio": 0.20014881938695908, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.012170064402744174, + "grad_norm": 0.07727162539958954, + "kept_token_count": 654.5484375, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9934427917003632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006557209696620703, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2869873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.10858764648435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23550, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.1997097223997116, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.012298346869647503, + "grad_norm": 0.12199007719755173, + "kept_token_count": 654.6078125, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9933260440826416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006673944741487503, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.58495178222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0340209960938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23560, + "valid_token_count": 817.9625 + }, + { + "actual_token_mask_ratio": 0.19976298063993453, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.012969151698052883, + "grad_norm": 0.10365894436836243, + "kept_token_count": 655.0296875, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9934767544269562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006523249018937349, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.4683013916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.34410705566404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23570, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.20021447092294692, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.011317389085888863, + "grad_norm": 0.11263928562402725, + "kept_token_count": 654.5578125, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934629499912262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537050846964121, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.72100524902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4495910644531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23580, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.19939396381378174, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.011542889382690191, + "grad_norm": 0.08331799507141113, + "kept_token_count": 655.19375, + "learning_rate": 3.592299907806881e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9930603802204132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939612235873938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5842651367187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.06171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23590, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.19963445216417314, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.01174774975515902, + "grad_norm": 0.07487451285123825, + "kept_token_count": 654.9484375, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9932060420513154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006793956179171801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.20696716308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4719696044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23600, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19924655854701995, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.011352814687415957, + "grad_norm": 0.10508260130882263, + "kept_token_count": 655.059375, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9932897090911865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710304785519839, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.89369812011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6816345214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23610, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.20024722814559937, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.013465230539441108, + "grad_norm": 0.10860567539930344, + "kept_token_count": 654.34375, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9933089792728425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066910235211253164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.42518920898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3948516845703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23620, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.19932459890842438, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.010917859058827162, + "grad_norm": 0.12444058060646057, + "kept_token_count": 655.3828125, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935266315937042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006473362632095814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.5114807128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4223205566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23630, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.19984101355075837, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.010674085887148976, + "grad_norm": 0.10645867884159088, + "kept_token_count": 654.7421875, + "learning_rate": 3.538791705098866e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933710873126984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006628910545259714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.4213562011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.56768188476565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20086710155010223, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.011512245424091816, + "grad_norm": 0.08812606334686279, + "kept_token_count": 653.846875, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9934065103530884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006593477167189121, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.2826721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9013244628906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23650, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.20001983195543288, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.01404965166002512, + "grad_norm": 0.09850054979324341, + "kept_token_count": 654.54375, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9937538862228393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006246133893728256, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.4002197265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23660, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.1994815081357956, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.011115829786285758, + "grad_norm": 0.08021323382854462, + "kept_token_count": 655.1375, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9935094594955445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00649054441601038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.720263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7856079101563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23670, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19956301748752595, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.010186556028202176, + "grad_norm": 0.09441522508859634, + "kept_token_count": 654.884375, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9928004443645477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007199560571461916, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2744873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7528961181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23680, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.20036431401968002, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.011369568482041358, + "grad_norm": 0.09458214044570923, + "kept_token_count": 654.4296875, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9932526648044586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747335847467184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.48077697753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9556427001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23690, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.19988815486431122, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.010748331341892481, + "grad_norm": 0.0877736285328865, + "kept_token_count": 654.53125, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9933143198490143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006685676891356707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.90460815429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8064025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23700, + "valid_token_count": 818.0453125 + }, + { + "actual_token_mask_ratio": 0.20058663338422775, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.010984532628208398, + "grad_norm": 0.08787655830383301, + "kept_token_count": 653.9421875, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9934838473796844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006516159046441317, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.16834716796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2213195800781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23710, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.20033522248268126, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.011985901556909085, + "grad_norm": 0.0972626656293869, + "kept_token_count": 654.3390625, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9930810630321503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006918940227478743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7676025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.22084350585936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23720, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.19982507526874543, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.011920371185988187, + "grad_norm": 0.08755073696374893, + "kept_token_count": 654.690625, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931593835353851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006840598676353693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5504547119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.99164123535155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23730, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19962380528450013, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.011773464921861886, + "grad_norm": 0.08696745336055756, + "kept_token_count": 655.04375, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9934422075748444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006557797733694315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.65148010253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.97718505859376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23740, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.20010486096143723, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.013126112287864089, + "grad_norm": 0.08906308561563492, + "kept_token_count": 654.74375, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.992991441488266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007008569780737162, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.00636901855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.50201110839845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23750, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.19999845921993256, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.012930544978007674, + "grad_norm": 0.10300066322088242, + "kept_token_count": 654.6078125, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.9934910476207733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065089289098978044, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.531005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5356018066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20019171386957169, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.011091614374890924, + "grad_norm": 0.10389890521764755, + "kept_token_count": 654.428125, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9927197158336639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007280275784432888, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1036834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23915710449216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23770, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.19999641925096512, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.011302984738722443, + "grad_norm": 0.10149861872196198, + "kept_token_count": 654.6078125, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936084508895874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006391549576073885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.93305053710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2550048828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23780, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20088173002004622, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.011270972108468414, + "grad_norm": 0.08176512271165848, + "kept_token_count": 653.7796875, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9930668354034424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006933159753680229, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8168640136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9247222900391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23790, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.1988864541053772, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.011311296513304115, + "grad_norm": 0.09247183054685593, + "kept_token_count": 655.35, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933274030685425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006672599632292986, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8743133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9750274658203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23800, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20031229704618453, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.012302793515846133, + "grad_norm": 0.1218089759349823, + "kept_token_count": 654.4859375, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9931979715824127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00680202329531312, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.77281188964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1438385009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23810, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20003654956817626, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.011871036933735013, + "grad_norm": 0.09496162831783295, + "kept_token_count": 654.4875, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931021630764008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006897841673344373, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1723114013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8308380126953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23820, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.19918332695961, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.013642409443855285, + "grad_norm": 0.10929661244153976, + "kept_token_count": 655.3, + "learning_rate": 3.338649462358145e-06, + "loss": 0.017, + "mgd_cos_sim": 0.9932745516300201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725456286221743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4442169189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3804565429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23830, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19977976828813554, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.012125726323574781, + "grad_norm": 0.10164272040128708, + "kept_token_count": 654.95625, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9926296353340149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00737036457285285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5115264892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4049468994141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23840, + "valid_token_count": 818.4671875 + }, + { + "actual_token_mask_ratio": 0.19998251050710678, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.013092098664492369, + "grad_norm": 0.14750449359416962, + "kept_token_count": 654.621875, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9929032981395721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007096701581031084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.56824645996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.98565673828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23850, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.19931265711784363, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012176774395629763, + "grad_norm": 0.09130433201789856, + "kept_token_count": 655.0671875, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933334112167358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006666587200015783, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8910675048828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9883270263672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23860, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.19976768493652344, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.010418729530647397, + "grad_norm": 0.11022075265645981, + "kept_token_count": 654.959375, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9935461699962616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006453822273761034, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.03302307128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5204345703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23870, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.19976565092802048, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.010133663052693009, + "grad_norm": 0.08928771317005157, + "kept_token_count": 654.81875, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9934080004692077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006592005118727684, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.67821350097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.02076416015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23880, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2004386827349663, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.01098466427065432, + "grad_norm": 0.09773960709571838, + "kept_token_count": 654.4953125, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9931950867176056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006804914586246014, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2551635742187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8615966796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23890, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.19986677765846253, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.012025863397866488, + "grad_norm": 0.08754733204841614, + "kept_token_count": 654.728125, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936412870883942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006358713936060667, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.4510772705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1823486328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23900, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.20084432065486907, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.010646424535661937, + "grad_norm": 0.07573229819536209, + "kept_token_count": 654.15, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9938431918621063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156800873577594, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1240997314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10675354003905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23910, + "valid_token_count": 818.5515625 + }, + { + "actual_token_mask_ratio": 0.1997755691409111, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.009522296534851194, + "grad_norm": 0.09230489283800125, + "kept_token_count": 655.0890625, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9933294236660004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00667057391256094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.30946044921876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2677062988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23920, + "valid_token_count": 818.6296875 + }, + { + "actual_token_mask_ratio": 0.19976319521665573, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.011352664791047573, + "grad_norm": 0.10623295605182648, + "kept_token_count": 654.878125, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9930864453315735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006913541164249182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.57237243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.77576904296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23930, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.19913950711488723, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.011560609564185143, + "grad_norm": 0.08537233620882034, + "kept_token_count": 655.4546875, + "learning_rate": 3.225110003817418e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9931600570678711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006839940324425697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.18121643066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.25710144042966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23940, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.20041040182113648, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.012827340373769402, + "grad_norm": 0.14829523861408234, + "kept_token_count": 654.5375, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9931388556957245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006861145701259374, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.45249328613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7712860107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23950, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19981937855482101, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.013730370718985797, + "grad_norm": 0.10446899384260178, + "kept_token_count": 655.040625, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9931362092494964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00686377976089716, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.13184814453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.76222229003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23960, + "valid_token_count": 818.61875 + }, + { + "actual_token_mask_ratio": 0.19987435042858123, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.013112161913886667, + "grad_norm": 0.367326945066452, + "kept_token_count": 654.6390625, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9931855082511902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006814483553171158, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.5229522705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.72754821777346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23970, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.1997713938355446, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.011099221045151353, + "grad_norm": 0.10266710817813873, + "kept_token_count": 654.7765625, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933422327041626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006657768227159977, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.7770660400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.53734130859374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23980, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.19999777376651764, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.011545070307329297, + "grad_norm": 0.10803297907114029, + "kept_token_count": 654.7234375, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9932747840881347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672521460801363, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.64617919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4508026123047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 23990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.19929558038711548, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.01055611139163375, + "grad_norm": 0.0962272435426712, + "kept_token_count": 655.190625, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9929220736026764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00707792267203331, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.2453155517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1189697265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24000, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20025139749050141, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.012952074687927962, + "grad_norm": 0.06924336403608322, + "kept_token_count": 654.1796875, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9933410584926605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066589491441845896, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.40062255859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4316650390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24010, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.19998577237129211, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.011776595097035169, + "grad_norm": 0.06790924817323685, + "kept_token_count": 654.728125, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933760643005372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006623940914869309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2020721435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3666656494141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24020, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19980866611003875, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.010825808765366674, + "grad_norm": 0.15203821659088135, + "kept_token_count": 654.871875, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9932821094989777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006717884168028832, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.13976135253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1724853515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24030, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.2001407489180565, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.009346127836033702, + "grad_norm": 0.09937665611505508, + "kept_token_count": 654.409375, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935392498970032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006460757646709681, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0636444091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9139465332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24040, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.200509712100029, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.011656866082921625, + "grad_norm": 0.13299791514873505, + "kept_token_count": 654.26875, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9929335474967956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007066458370536566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3718597412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4513000488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24050, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.200143001973629, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.011009047878906131, + "grad_norm": 0.1166786253452301, + "kept_token_count": 654.296875, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9933780312538147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006621963437646627, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.63804931640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1964477539062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24060, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.19981296807527543, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.011887204926460981, + "grad_norm": 0.09538991749286652, + "kept_token_count": 654.7578125, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9930732846260071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006926714535802602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1917236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.36871643066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24070, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.1996139869093895, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.011369638424366713, + "grad_norm": 0.12132071703672409, + "kept_token_count": 655.0859375, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9938101530075073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061898637562990185, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0523162841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6444122314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24080, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.20053186565637587, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.011125029623508453, + "grad_norm": 0.10321643948554993, + "kept_token_count": 653.95, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9931582391262055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006841771584004164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4688659667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6623199462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24090, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.2002166762948036, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.01186708202585578, + "grad_norm": 0.1348947435617447, + "kept_token_count": 654.1625, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933869779109955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613022927194834, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.44994201660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.80174255371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24100, + "valid_token_count": 817.921875 + }, + { + "actual_token_mask_ratio": 0.1995126336812973, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.01002445351332426, + "grad_norm": 0.07850970327854156, + "kept_token_count": 655.115625, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9934859216213227, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00651409849524498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.6234893798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3161102294922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24110, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.1999232977628708, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.010393619211390615, + "grad_norm": 0.1280880570411682, + "kept_token_count": 654.4078125, + "learning_rate": 3.043059933362236e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9928552150726319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007144780643284321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.38829345703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8162353515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24120, + "valid_token_count": 817.93125 + }, + { + "actual_token_mask_ratio": 0.1996745303273201, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.012176167406141757, + "grad_norm": 0.09874571859836578, + "kept_token_count": 654.7953125, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9932254791259766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067745066247880455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.415576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0902801513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24130, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20007685124874114, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.010321111464872957, + "grad_norm": 0.10501005500555038, + "kept_token_count": 654.5671875, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9928001344203949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071998628787696365, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.82939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3133544921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24140, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19990600347518922, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.010402194503694773, + "grad_norm": 0.06938599795103073, + "kept_token_count": 654.53125, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9932132422924042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006786762457340956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1881561279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7285186767578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24150, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.19982996433973313, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.012784654460847377, + "grad_norm": 0.07763494551181793, + "kept_token_count": 654.5828125, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933773636817932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006622626632452011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 448.53899536132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.81103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24160, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.1999713346362114, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.011709416331723333, + "grad_norm": 0.10159855335950851, + "kept_token_count": 654.7828125, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9935362756252288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006463721115142107, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.35552673339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1114837646484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24170, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.19887375831604004, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.012425945466384291, + "grad_norm": 0.10441168397665024, + "kept_token_count": 655.45, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9930929899215698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906998343765736, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.79395751953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4188934326172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.20045881420373918, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.009886124730110168, + "grad_norm": 0.07633458077907562, + "kept_token_count": 654.2921875, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9931108236312867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889188382774592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4780792236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0803955078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24190, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19939401149749755, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.012851141393184662, + "grad_norm": 0.11157430708408356, + "kept_token_count": 655.140625, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9925833940505981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007416605856269598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0864685058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84942321777345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24200, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2006174236536026, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.012321430491283536, + "grad_norm": 0.13247521221637726, + "kept_token_count": 654.309375, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9929288625717163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007071137893944979, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9770812988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0906616210938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24210, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19985745698213578, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.011846359865739942, + "grad_norm": 0.13965967297554016, + "kept_token_count": 654.7203125, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931116819381713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006888323184102774, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.9423400878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6214233398438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24220, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.20094146728515624, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.009977536322548986, + "grad_norm": 0.07186673581600189, + "kept_token_count": 653.6359375, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933830201625824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00661698579788208, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.5454864501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.44963989257815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24230, + "valid_token_count": 818.0078125 + }, + { + "actual_token_mask_ratio": 0.19949049651622772, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.010221230238676072, + "grad_norm": 0.07820218801498413, + "kept_token_count": 655.1203125, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9934963762760163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065036403015255925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2416168212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2557373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24240, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.19970690608024597, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.01121991639956832, + "grad_norm": 0.08939018100500107, + "kept_token_count": 654.909375, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9930118918418884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006988109927624464, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.13978881835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8467529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24250, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.19926002621650696, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.010653486009687185, + "grad_norm": 0.08269494026899338, + "kept_token_count": 655.1875, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9936454594135284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006354539841413498, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8586669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9472320556641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24260, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.19996360540390015, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.011540987715125083, + "grad_norm": 0.073182612657547, + "kept_token_count": 654.8578125, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9931969463825225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006803058180958032, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1525512695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2594299316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24270, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.19943133294582366, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.010808366257697344, + "grad_norm": 0.08749187737703323, + "kept_token_count": 655.2875, + "learning_rate": 2.885187168570864e-06, + "loss": 0.014, + "mgd_cos_sim": 0.993552052974701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064479528926312925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.9931304931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1466125488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24280, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.1993718758225441, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.013137290813028812, + "grad_norm": 0.10437827557325363, + "kept_token_count": 655.165625, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0167, + "mgd_cos_sim": 0.9927776098251343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00722240200266242, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.1177215576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2646820068359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24290, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.19983685463666917, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.01188198458403349, + "grad_norm": 0.09351220726966858, + "kept_token_count": 654.628125, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9936427295207977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006357267964631319, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.5252410888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.55934448242186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24300, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.19945070743560792, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.01297214082442224, + "grad_norm": 0.09946410357952118, + "kept_token_count": 655.065625, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.993812745809555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00618726210668683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4659790039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.55100708007814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24310, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.2002575308084488, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.012453659065067768, + "grad_norm": 0.10914584249258041, + "kept_token_count": 654.234375, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9933280706405639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00667194053530693, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.87050476074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.190966796875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24320, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20068760663270951, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.010548653500154614, + "grad_norm": 0.09151139110326767, + "kept_token_count": 653.7359375, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9935555875301361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444429792463779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.0257171630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.51091918945315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24330, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.2007073849439621, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.01155822235159576, + "grad_norm": 0.08153973519802094, + "kept_token_count": 654.0125, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9938227593898773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00617723474279046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.3446441650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5141571044922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24340, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.19970685839653016, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.012197411013767123, + "grad_norm": 0.09523621201515198, + "kept_token_count": 654.8703125, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9929520964622498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007047902047634125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5937164306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6045623779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24350, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19925426691770554, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.011853991588577629, + "grad_norm": 0.09779732674360275, + "kept_token_count": 655.2046875, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9935876548290252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006412335112690926, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.05230712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.28795166015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24360, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2002100244164467, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.011114687751978635, + "grad_norm": 0.09170170873403549, + "kept_token_count": 654.3203125, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933810949325561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006618909630924463, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.33448486328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8682342529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24370, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.19957161992788314, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.010314442310482263, + "grad_norm": 0.08495645970106125, + "kept_token_count": 654.8078125, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9934853076934814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006514697801321745, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0587188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5100891113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24380, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.2001221463084221, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.011057974910363555, + "grad_norm": 0.13119155168533325, + "kept_token_count": 654.646875, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9932770073413849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006722982414066792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4781219482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.80967102050784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24390, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20022224336862565, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.011992053035646677, + "grad_norm": 0.14190925657749176, + "kept_token_count": 654.375, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9931684970855713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006831479538232088, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9707824707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0919586181641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24400, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.20033118277788162, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.013534731231629848, + "grad_norm": 0.1141134575009346, + "kept_token_count": 654.25625, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9933486938476562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00665130540728569, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7577819824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9740844726563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24410, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.19992415457963944, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.01088049216195941, + "grad_norm": 0.08745724707841873, + "kept_token_count": 654.4453125, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9932036221027374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006796377524733543, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2004364013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3014190673828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24420, + "valid_token_count": 817.9796875 + }, + { + "actual_token_mask_ratio": 0.19996873438358306, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.012853768561035395, + "grad_norm": 0.12050604075193405, + "kept_token_count": 654.8296875, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.993265426158905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067345801740884784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3898162841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.27113952636716, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24430, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.19980005472898482, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.012231360096484423, + "grad_norm": 0.1069214791059494, + "kept_token_count": 654.578125, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9934111773967743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588824559003115, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8059967041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.84311828613284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24440, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.1999540627002716, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.015552659332752228, + "grad_norm": 0.09808085113763809, + "kept_token_count": 654.753125, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0189, + "mgd_cos_sim": 0.9932492792606353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006750733498483897, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.23760681152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0293304443359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24450, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2002352088689804, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.008651677938178181, + "grad_norm": 0.09165368229150772, + "kept_token_count": 654.1140625, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9935959577560425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006404039822518826, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.27149658203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.47265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24460, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.2002451792359352, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.009412154462188483, + "grad_norm": 0.07305163890123367, + "kept_token_count": 654.3296875, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937375068664551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006262481771409511, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.1902313232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1028625488281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24470, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19900065958499907, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.012521661398932338, + "grad_norm": 0.09721458703279495, + "kept_token_count": 655.50625, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9935514748096466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448521837592125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.6842895507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9410034179688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24480, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.19977298527956008, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.009695344138890505, + "grad_norm": 0.07152125984430313, + "kept_token_count": 654.71875, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9935967087745666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006403287407010794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.3644561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6274688720703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24490, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.19846176058053971, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.012631246214732527, + "grad_norm": 0.07232837378978729, + "kept_token_count": 655.8328125, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932991862297058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006700795982033015, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.2973205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.21669006347656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24500, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2005906105041504, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.011767596937716007, + "grad_norm": 0.09699012339115143, + "kept_token_count": 654.21875, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9936466872692108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006353306677192449, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.56483459472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2813262939453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24510, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.19968635588884354, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.010555628407746553, + "grad_norm": 0.0903027355670929, + "kept_token_count": 655.08125, + "learning_rate": 2.655458772030739e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9931766986846924, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006823298055678606, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5148864746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1277648925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24520, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.1999206781387329, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.010602761059999466, + "grad_norm": 0.07683645188808441, + "kept_token_count": 654.5875, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9936479687690735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006352028343826532, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.42698059082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.14954528808596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2002841964364052, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.011357677076011896, + "grad_norm": 0.08067167550325394, + "kept_token_count": 654.575, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936501622200012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006349834147840738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2184234619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4932525634766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24540, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.19908345937728883, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.011338277300819755, + "grad_norm": 0.06708022952079773, + "kept_token_count": 655.3828125, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931916177272797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068083880469202995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6633728027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4646728515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24550, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19936595112085342, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.011461298773065209, + "grad_norm": 0.10910968482494354, + "kept_token_count": 654.99375, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9932685315608978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006731477845460176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5185150146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.26634826660154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24560, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.20079025328159333, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.01061163004487753, + "grad_norm": 0.10709603130817413, + "kept_token_count": 654.1046875, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9934905886650085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006509398203343153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.49501342773436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7968444824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24570, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.19890504479408264, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.011229859618470072, + "grad_norm": 0.08797964453697205, + "kept_token_count": 655.58125, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933826088905334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617399118840695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.03713073730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.24847717285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24580, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.20020794719457627, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.01124505428597331, + "grad_norm": 0.13977044820785522, + "kept_token_count": 654.4015625, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9929075598716736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092428579926491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5335235595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7252258300781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24590, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.19929984956979752, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.011552422773092985, + "grad_norm": 0.07956647127866745, + "kept_token_count": 655.25625, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9935290813446045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006470919586718082, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.6409637451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1565338134766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24600, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19988452792167663, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.01027514822781086, + "grad_norm": 0.09254920482635498, + "kept_token_count": 654.753125, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936337411403656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006366265751421451, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.24970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0710693359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24610, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2003257617354393, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.011255700839683414, + "grad_norm": 0.09620833396911621, + "kept_token_count": 654.18125, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933824062347412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617576349526644, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.012255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.55698852539064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24620, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.20019933581352234, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.010599062824621797, + "grad_norm": 0.07981529086828232, + "kept_token_count": 654.3953125, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.993584257364273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064157385379076, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8340240478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7943420410156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24630, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.19984588623046876, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.010415916517376899, + "grad_norm": 0.11843354254961014, + "kept_token_count": 654.8359375, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9934320271015167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006567975971847773, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1694610595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.54420166015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24640, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.20014804601669312, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.009491701470687985, + "grad_norm": 0.0717289075255394, + "kept_token_count": 654.2078125, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938465833663941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061534280888736245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.67344665527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.482568359375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24650, + "valid_token_count": 817.9125 + }, + { + "actual_token_mask_ratio": 0.19978834837675094, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.011916695581749081, + "grad_norm": 0.08834851533174515, + "kept_token_count": 654.7890625, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9934539079666138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006546096038073301, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.49871520996095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.39012451171874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24660, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.20088550746440886, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.010262386268004775, + "grad_norm": 0.09711012989282608, + "kept_token_count": 653.8234375, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935129284858704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006487070955336094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1003173828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9853454589844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24670, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.19962267577648163, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.010273104207590223, + "grad_norm": 0.14179293811321259, + "kept_token_count": 654.7921875, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9930181384086609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006981862895190716, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.83123779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.17982788085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24680, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.20054195821285248, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.012527674436569214, + "grad_norm": 0.10448943078517914, + "kept_token_count": 654.4125, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9928238749504089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007176121696829796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.356005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9410369873047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24690, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.20005581378936768, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.010828395467251539, + "grad_norm": 0.11871839314699173, + "kept_token_count": 654.4640625, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934969186782837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006503081880509853, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.12074584960936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58958740234374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24700, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19977044463157653, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.012964093126356602, + "grad_norm": 0.10855177789926529, + "kept_token_count": 654.625, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9929243862628937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007075630687177181, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1333709716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8080627441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24710, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.1998193547129631, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010087493155151606, + "grad_norm": 0.06994166970252991, + "kept_token_count": 654.621875, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9931274354457855, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006872565671801567, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.0040679931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.30707397460935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.20008091181516646, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.010954071721062063, + "grad_norm": 0.11080262809991837, + "kept_token_count": 654.6234375, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9932361841201782, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006763819977641106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1345275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0850463867188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24730, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.1996372312307358, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.01214795671403408, + "grad_norm": 0.08847244083881378, + "kept_token_count": 655.203125, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.993406069278717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006593945622444153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0007293701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6918518066406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24740, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.20110715925693512, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.011228215880692005, + "grad_norm": 0.1063840389251709, + "kept_token_count": 654.0671875, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9927990972995758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007200897391885519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.597802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.34031066894534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24750, + "valid_token_count": 818.7125 + }, + { + "actual_token_mask_ratio": 0.20001992136240004, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.010608017211779952, + "grad_norm": 0.18883377313613892, + "kept_token_count": 654.409375, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9931042730808258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006895721144974232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8331573486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.47092895507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24760, + "valid_token_count": 818.025 + }, + { + "actual_token_mask_ratio": 0.19992893934249878, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.012377881025895477, + "grad_norm": 0.10024509578943253, + "kept_token_count": 654.8171875, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9932457625865936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006754228100180626, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5825439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.84571228027346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24770, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.1998369425535202, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.01047824528068304, + "grad_norm": 0.07206852734088898, + "kept_token_count": 654.9453125, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9936187565326691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381251011043787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1532409667969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.18052368164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24780, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.20022087097167968, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.01246101465076208, + "grad_norm": 0.1000216081738472, + "kept_token_count": 654.540625, + "learning_rate": 2.407348847176735e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9929692029953003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007030785363167524, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1682891845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2480224609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24790, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.19948523938655854, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.011390605242922902, + "grad_norm": 0.10121262818574905, + "kept_token_count": 655.1203125, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.993300324678421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006699671503156424, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1815673828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.94054260253904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24800, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.19958456307649614, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.011639170674607157, + "grad_norm": 0.11303707212209702, + "kept_token_count": 655.1078125, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9936237692832947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006376238912343979, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2106658935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8092956542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24810, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.20029521435499192, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012256380869075656, + "grad_norm": 0.12697136402130127, + "kept_token_count": 654.540625, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993148821592331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006851167418062687, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5103057861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57342529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24820, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20052223056554794, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.01165944947861135, + "grad_norm": 0.10465189069509506, + "kept_token_count": 654.0765625, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9934440791606903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006555904354900122, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.7292144775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0227081298828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24830, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.2000589594244957, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.01067691808566451, + "grad_norm": 0.0856657400727272, + "kept_token_count": 654.4625, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9933605909347534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639418099075556, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7544403076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.40875854492185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24840, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19972781836986542, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.010956531995907426, + "grad_norm": 0.10280224680900574, + "kept_token_count": 654.734375, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935083448886871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006491655204445124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2845520019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48182067871096, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24850, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.19947388172149658, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.016180542949587106, + "grad_norm": 0.09884093701839447, + "kept_token_count": 655.2734375, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0197, + "mgd_cos_sim": 0.9929359674453735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070640261285007, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5750274658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84535217285156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24860, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.19969796687364577, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.011257418384775519, + "grad_norm": 0.13404372334480286, + "kept_token_count": 655.0046875, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931894063949585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068105868063867096, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.41180725097655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.28134765625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24870, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.20017440915107726, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.011766067938879132, + "grad_norm": 0.11830078810453415, + "kept_token_count": 654.41875, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9933976590633392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006602336559444666, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3908996582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22735595703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24880, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.20050135105848313, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.01154428874142468, + "grad_norm": 0.10907993465662003, + "kept_token_count": 654.3609375, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9930017411708831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069982543587684635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3969207763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.73267822265626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24890, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.20026755183935166, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.01105523924343288, + "grad_norm": 0.08939378708600998, + "kept_token_count": 654.590625, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9929233193397522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007076685223728419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3370300292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.54291381835935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24900, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.20112141370773315, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.01277183904312551, + "grad_norm": 0.0919058620929718, + "kept_token_count": 653.8109375, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9935265958309174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006473411154001951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.28850708007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0027618408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24910, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.19951080679893493, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.012583768228068948, + "grad_norm": 0.12848128378391266, + "kept_token_count": 655.171875, + "learning_rate": 2.291853375705327e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932525813579559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747404579073191, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.44353637695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8258422851562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24920, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.200521819293499, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.012407276453450322, + "grad_norm": 0.09935998171567917, + "kept_token_count": 654.34375, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9934211790561676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006578819081187248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0351898193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9631317138672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24930, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.200182543694973, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.01214006645604968, + "grad_norm": 0.13264675438404083, + "kept_token_count": 654.3140625, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9933909118175507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066091052256524565, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.48411254882814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.35158386230466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24940, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.1993873655796051, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.01011853339150548, + "grad_norm": 0.09131729602813721, + "kept_token_count": 655.0390625, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9935045182704926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006495483405888081, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.409228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7068145751953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24950, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19936281740665435, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.012208744790405035, + "grad_norm": 0.095608189702034, + "kept_token_count": 655.1, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9932607233524322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00673926891759038, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9216766357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2262817382813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24960, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19910238832235336, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.01198625350371003, + "grad_norm": 0.09143069386482239, + "kept_token_count": 655.3, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930054605007171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006994541734457016, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.75625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.066748046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24970, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19903960227966308, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.010583938146010042, + "grad_norm": 0.08645550161600113, + "kept_token_count": 655.215625, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9938557624816895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006144245062023402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.06443176269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4009307861328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24980, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.19979672133922577, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.0105069974437356, + "grad_norm": 0.0793650671839714, + "kept_token_count": 654.5953125, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9934809386730195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065190671943128106, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.34327697753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2403167724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 24990, + "valid_token_count": 818.0375 + }, + { + "actual_token_mask_ratio": 0.2004328638315201, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.014134641084820033, + "grad_norm": 0.09652568399906158, + "kept_token_count": 654.359375, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9938020884990693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006197909079492092, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.9943603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2839416503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25000, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.1994830146431923, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.010945017542690038, + "grad_norm": 0.11305710673332214, + "kept_token_count": 654.9640625, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9932126462459564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006787351612001657, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.39146423339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.89446411132815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25010, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.19937629550695418, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.010438640369102359, + "grad_norm": 0.12421669811010361, + "kept_token_count": 655.0890625, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9932625889778137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067374113947153095, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.52366027832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7899597167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25020, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.199517560005188, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.012206973740831017, + "grad_norm": 0.07275831699371338, + "kept_token_count": 654.6828125, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.992739337682724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007260655425488949, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.60113830566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0101745605469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25030, + "valid_token_count": 817.859375 + }, + { + "actual_token_mask_ratio": 0.20011122077703475, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.013091412512585521, + "grad_norm": 0.0884433463215828, + "kept_token_count": 654.5375, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9934408068656921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006559191271662712, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6580535888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1210601806641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25040, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.20060070902109145, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.010283909877762199, + "grad_norm": 0.08524999022483826, + "kept_token_count": 653.9625, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936205685138703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006379428692162037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5542816162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.74107666015624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25050, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.19957427978515624, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.013002697797492146, + "grad_norm": 0.0871722623705864, + "kept_token_count": 654.725, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9933065295219421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006693484354764223, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5187622070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.154541015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25060, + "valid_token_count": 817.9703125 + }, + { + "actual_token_mask_ratio": 0.1993117392063141, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.010954737896099687, + "grad_norm": 0.13556252419948578, + "kept_token_count": 655.084375, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9928780019283294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007121999282389879, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5348327636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.98740234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25070, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.1996639147400856, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.01403487273491919, + "grad_norm": 0.0891500860452652, + "kept_token_count": 654.7796875, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9936186730861664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381321419030428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.46279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6518310546875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.19957004189491273, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010757987620308995, + "grad_norm": 0.10080907493829727, + "kept_token_count": 655.0296875, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934224903583526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006577518675476313, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2095581054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5851135253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25090, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.19949106872081757, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.010525357583537698, + "grad_norm": 0.11829816550016403, + "kept_token_count": 655.078125, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9933318972587586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006668103300035, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6508087158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.11345825195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25100, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20014996379613875, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.01258233580738306, + "grad_norm": 0.1107085794210434, + "kept_token_count": 654.453125, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9935463309288025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006453694310039282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.39182739257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57021484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25110, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19986820369958877, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.009787003044039011, + "grad_norm": 0.10318335145711899, + "kept_token_count": 654.625, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9933465361595154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653483398258686, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3781463623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0673431396484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25120, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.20052581578493117, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.012625405052676796, + "grad_norm": 0.1489371359348297, + "kept_token_count": 654.0921875, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9926616191864014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073383752256631855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.37860412597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.24176025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25130, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.2002527743577957, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.01243139998987317, + "grad_norm": 0.07907121628522873, + "kept_token_count": 654.1390625, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9933291733264923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006670842971652746, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1759796142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3702819824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25140, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.19936449974775314, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.010475032683461904, + "grad_norm": 0.0727275013923645, + "kept_token_count": 655.225, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9932441473007202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006755854096263647, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5342956542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2251403808594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25150, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2005883514881134, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.011092192959040403, + "grad_norm": 0.10075032711029053, + "kept_token_count": 654.475, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935730040073395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064269925467669966, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8981536865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.183154296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25160, + "valid_token_count": 818.696875 + }, + { + "actual_token_mask_ratio": 0.1997094765305519, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.010499023459851741, + "grad_norm": 0.10909813642501831, + "kept_token_count": 654.61875, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9929923951625824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007007612474262714, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4453918457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.16315002441405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25170, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.20100706815719604, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.011744381999596954, + "grad_norm": 0.14574946463108063, + "kept_token_count": 653.975, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9934343338012696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006565669365227222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.59617614746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6377227783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25180, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.19979889690876007, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.011416332889348268, + "grad_norm": 0.09403248131275177, + "kept_token_count": 654.6484375, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9937659919261932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00623400267213583, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.8800994873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1486114501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25190, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.200139819085598, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.011383552243933082, + "grad_norm": 0.06710511445999146, + "kept_token_count": 654.634375, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9930905699729919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006909428909420967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.0124114990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.20859680175784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25200, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.19988456964492798, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.011983741074800491, + "grad_norm": 0.09518265724182129, + "kept_token_count": 654.6671875, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9932710647583007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067289381287992, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.6235015869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.38612670898436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25210, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2009625107049942, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.011392841814085841, + "grad_norm": 0.1524025797843933, + "kept_token_count": 653.628125, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9935442447662354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00645575001835823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.9704162597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2218475341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25220, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.19941070973873137, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.011273518484085799, + "grad_norm": 0.07196073234081268, + "kept_token_count": 654.859375, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9930183291435242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006981663405895233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.69814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5174224853516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25230, + "valid_token_count": 817.9734375 + }, + { + "actual_token_mask_ratio": 0.19963372647762298, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.010338816279545426, + "grad_norm": 0.05813629925251007, + "kept_token_count": 654.7859375, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9938437879085541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156205479055643, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3280395507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9085723876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25240, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20000761300325393, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.011930654942989349, + "grad_norm": 0.09365717321634293, + "kept_token_count": 654.5984375, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933667361736298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066332691349089146, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0604644775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5339782714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25250, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19959118217229843, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.011447414243593811, + "grad_norm": 0.08245731145143509, + "kept_token_count": 654.703125, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9930008351802826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006999160721898079, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.14228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.54652709960936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25260, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19965157955884932, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.012787457369267941, + "grad_norm": 0.12401469051837921, + "kept_token_count": 655.0421875, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9933717012405395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006628302205353975, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.094970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.695166015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25270, + "valid_token_count": 818.446875 + }, + { + "actual_token_mask_ratio": 0.1998934715986252, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.010518930247053504, + "grad_norm": 0.07626610994338989, + "kept_token_count": 654.865625, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9927755534648895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224443182349205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.86842956542966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9238555908203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25280, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.200363452732563, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.010392763139680027, + "grad_norm": 0.06985345482826233, + "kept_token_count": 653.9421875, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9934411764144897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006558832246810198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8172271728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.209814453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25290, + "valid_token_count": 817.7984375 + }, + { + "actual_token_mask_ratio": 0.2004491224884987, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.01071109427139163, + "grad_norm": 0.0710805281996727, + "kept_token_count": 654.178125, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934504270553589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006549570523202419, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8321807861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62278747558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.200197072327137, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.011652810359373688, + "grad_norm": 0.09963113814592361, + "kept_token_count": 654.7359375, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9932887375354766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006711254827678204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.03940734863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2334350585937, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25310, + "valid_token_count": 818.6234375 + }, + { + "actual_token_mask_ratio": 0.200080107152462, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.010023332526907324, + "grad_norm": 0.08555632084608078, + "kept_token_count": 654.721875, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9938295900821685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006170406844466925, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.4549560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9796630859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25320, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.19965653419494628, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.012087075831368565, + "grad_norm": 0.11994819343090057, + "kept_token_count": 655.0640625, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0154, + "mgd_cos_sim": 0.9934247314929963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006575271114706993, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.39488220214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6202911376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25330, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.1997939169406891, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.01169976145029068, + "grad_norm": 0.10366465896368027, + "kept_token_count": 654.4421875, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9933615684509277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066384184174239635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.42669372558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0080322265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25340, + "valid_token_count": 817.8421875 + }, + { + "actual_token_mask_ratio": 0.19996478855609895, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012043134775012732, + "grad_norm": 0.07300836592912674, + "kept_token_count": 654.5859375, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936262190341949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006373780872672796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.244482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.27064208984376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25350, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19950286149978638, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.010318576358258724, + "grad_norm": 0.07139793783426285, + "kept_token_count": 655.1015625, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.993050616979599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069493834860622885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.58604125976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.67469787597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25360, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.1998506009578705, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.01117852465249598, + "grad_norm": 0.11802570521831512, + "kept_token_count": 654.728125, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9935235977172852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006476408708840609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.1150634765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62403564453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25370, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.19933271110057832, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.011259068129584193, + "grad_norm": 0.08432400971651077, + "kept_token_count": 655.1734375, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931307196617126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006869277078658342, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.3165222167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2333282470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25380, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.19915515035390854, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.011398973548784852, + "grad_norm": 0.10414978861808777, + "kept_token_count": 655.2015625, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9929065763950348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070934263989329335, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9658508300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.00049743652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25390, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.1996890962123871, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.009974471013993024, + "grad_norm": 0.09138491004705429, + "kept_token_count": 654.9140625, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933870136737823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612993311136961, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.1329711914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5679138183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25400, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.20021240264177323, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.01228692550212145, + "grad_norm": 0.10055297613143921, + "kept_token_count": 654.43125, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9929871618747711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007012833282351494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.4095062255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.66302490234375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25410, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.19977053701877595, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.012785605806857347, + "grad_norm": 0.0811370313167572, + "kept_token_count": 655.1046875, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9937199234962464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628007659688592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.79265747070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3684997558594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25420, + "valid_token_count": 818.6453125 + }, + { + "actual_token_mask_ratio": 0.19999137818813323, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.010740092676132917, + "grad_norm": 0.08575203269720078, + "kept_token_count": 654.678125, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934933900833129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006506617739796638, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0906097412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8316375732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25430, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19987040609121323, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.009729220205917955, + "grad_norm": 0.09555843472480774, + "kept_token_count": 654.621875, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9931678175926208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006832181941717863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.6873382568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4549102783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25440, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.20005298107862474, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.011914830841124057, + "grad_norm": 0.09095192700624466, + "kept_token_count": 654.9328125, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9932765901088715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006723407935351133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.7316650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6945861816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25450, + "valid_token_count": 818.721875 + }, + { + "actual_token_mask_ratio": 0.19996987134218216, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.009567250171676278, + "grad_norm": 0.11582957953214645, + "kept_token_count": 654.6359375, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9934818983078003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00651809386909008, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.30353393554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3843017578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25460, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19936060905456543, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.012303171586245299, + "grad_norm": 0.08939110487699509, + "kept_token_count": 655.1046875, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.993192982673645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006807016395032406, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0517059326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5556243896484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25470, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2009938582777977, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.012972161546349526, + "grad_norm": 0.0885983482003212, + "kept_token_count": 653.6890625, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9931983947753906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006801613885909319, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6525177001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.523974609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25480, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.20024262219667435, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.011325920699164271, + "grad_norm": 0.0726676881313324, + "kept_token_count": 654.1859375, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9933595597743988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006640438549220562, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3739715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86549377441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25490, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19975070506334305, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.012517519667744637, + "grad_norm": 0.12766480445861816, + "kept_token_count": 654.9703125, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9932323276996613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00676767760887742, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.6733001708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.71753845214846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25500, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2006607323884964, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.011180265806615353, + "grad_norm": 0.11381243914365768, + "kept_token_count": 654.146875, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9936209678649902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637902831658721, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.75331115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1944915771484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25510, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.19956947565078736, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.011361348396167158, + "grad_norm": 0.08092239499092102, + "kept_token_count": 655.2171875, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9932264447212219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067735609598457815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6619049072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1902282714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25520, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.19971650540828706, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.012734120339155197, + "grad_norm": 0.09357281029224396, + "kept_token_count": 654.8140625, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9932583630084991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006741644907742738, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0851715087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8964324951172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25530, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.20040150582790375, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.011529142502695322, + "grad_norm": 0.20237413048744202, + "kept_token_count": 654.3484375, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9933079540729522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006692043971270323, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8188537597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5392150878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25540, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.19994499087333678, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012158274604007601, + "grad_norm": 0.08841854333877563, + "kept_token_count": 654.6375, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9929896712303161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007010349165648222, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.31899108886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.01386108398435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25550, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.1983189269900322, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.012286868318915367, + "grad_norm": 0.12092626839876175, + "kept_token_count": 656.165625, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9930892288684845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069107779301702974, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.11785888671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.77552185058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25560, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.20023048520088196, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.012608143966645002, + "grad_norm": 0.11259602755308151, + "kept_token_count": 654.3828125, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9933443486690521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006655648909509182, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0089111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.07493286132814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25570, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20014152228832244, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.011762889241799713, + "grad_norm": 0.1263294816017151, + "kept_token_count": 654.6703125, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931971967220307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006802806165069342, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.52373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.71009826660156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25580, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.20038872361183166, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.012930094543844462, + "grad_norm": 0.09506965428590775, + "kept_token_count": 654.31875, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932507753372193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006749226991087198, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6368469238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.90782165527344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25590, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.20015435814857482, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.012167762313038111, + "grad_norm": 0.10155052691698074, + "kept_token_count": 654.7625, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9932762205600738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672376649454236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6885040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4590728759766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25600, + "valid_token_count": 818.6109375 + }, + { + "actual_token_mask_ratio": 0.20027138888835908, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.012059070309624076, + "grad_norm": 0.1176363006234169, + "kept_token_count": 654.2140625, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9930658221244812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006934184860438108, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.24472961425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7601654052734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25610, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.19979128539562224, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.01036095074377954, + "grad_norm": 0.10219991952180862, + "kept_token_count": 654.8921875, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9935138523578644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006486153975129127, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4682312011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1263916015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25620, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.20070840120315553, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.012193651963025331, + "grad_norm": 0.15018023550510406, + "kept_token_count": 654.2625, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9934727907180786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006527218967676163, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.700830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68468017578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25630, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.19897520840167998, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.01003147978335619, + "grad_norm": 0.08467146754264832, + "kept_token_count": 655.2265625, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9932607233524322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00673928102478385, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5613250732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58843994140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25640, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.20082453191280364, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.012055725511163472, + "grad_norm": 0.10993734747171402, + "kept_token_count": 653.9171875, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9929306983947754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007069308776408434, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.14207763671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9158447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25650, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.19937975853681564, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.010100450785830616, + "grad_norm": 0.08412753790616989, + "kept_token_count": 655.2015625, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9930929958820343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00690700588747859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.50425720214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.39849243164065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25660, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.20026310980319978, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.01172182196751237, + "grad_norm": 0.1416972279548645, + "kept_token_count": 654.45, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9936324000358582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006367611885070801, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.04897155761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1651916503906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25670, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20064706951379777, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.012984797311946749, + "grad_norm": 0.1154719889163971, + "kept_token_count": 653.865625, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.9932031273841858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067968728020787236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.98668518066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4006713867187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25680, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.20042671263217926, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.011656327545642853, + "grad_norm": 0.11096487939357758, + "kept_token_count": 654.4921875, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9929134309291839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007086559291929006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.46837158203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2349151611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25690, + "valid_token_count": 818.55625 + }, + { + "actual_token_mask_ratio": 0.19947833716869354, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.011335349828004836, + "grad_norm": 0.07786674052476883, + "kept_token_count": 654.965625, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9933461308479309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066538708284497265, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.6235046386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7459350585938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25700, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.19991210401058196, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.011877339147031308, + "grad_norm": 0.09513262659311295, + "kept_token_count": 654.3859375, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9932608723640441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006739125214517116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1783813476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.46070251464846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25710, + "valid_token_count": 817.8875 + }, + { + "actual_token_mask_ratio": 0.19993535578250884, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.011030399706214666, + "grad_norm": 0.10349453240633011, + "kept_token_count": 654.859375, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934201717376709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065798373892903324, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.92661743164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3312713623047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25720, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.20059546530246736, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.010332000255584717, + "grad_norm": 0.07957827299833298, + "kept_token_count": 654.015625, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9934171795845032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065828350372612475, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.3074127197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4800628662109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25730, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.1996346667408943, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.012547039985656738, + "grad_norm": 0.10281657427549362, + "kept_token_count": 654.728125, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9932483553886413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006751645356416702, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.00840148925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4967010498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25740, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20139315575361252, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.010319594340398908, + "grad_norm": 0.09067660570144653, + "kept_token_count": 653.325, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9930043995380402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069956046529114245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.66213073730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22808227539065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25750, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.2002541348338127, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.010810429649427532, + "grad_norm": 0.10892525315284729, + "kept_token_count": 654.4453125, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9931575775146484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068424297496676445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3747222900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3187652587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25760, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.20075372606515884, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.011449310090392828, + "grad_norm": 0.08854036033153534, + "kept_token_count": 653.9796875, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9932352364063263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067647542804479596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6454376220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1518951416016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.20001859962940216, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.010730368783697486, + "grad_norm": 0.10114017874002457, + "kept_token_count": 654.5265625, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934476792812348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552316620945931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.02391662597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2139953613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25780, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.20030016899108888, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.013439402263611556, + "grad_norm": 0.10350124537944794, + "kept_token_count": 654.253125, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.993156498670578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006843510642647743, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.29576416015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.99009399414064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25790, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20020368993282317, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.011253387341275812, + "grad_norm": 0.08392360061407089, + "kept_token_count": 654.2453125, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9931747317314148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068252657540142534, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.14226684570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3665466308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25800, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.19848013073205947, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.010682072909548878, + "grad_norm": 0.10117169469594955, + "kept_token_count": 655.90625, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9935238838195801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064761144109070305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.5639221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5529296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25810, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20015281438827515, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.011078642355278135, + "grad_norm": 0.07939968258142471, + "kept_token_count": 654.425, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933996319770813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00660037100315094, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.15338745117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.03492736816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25820, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.20015069991350173, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.010439090896397829, + "grad_norm": 0.057792212814092636, + "kept_token_count": 654.6390625, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9928732216358185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00712678087875247, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.57704772949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0268859863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25830, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.19905658066272736, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.012335891462862491, + "grad_norm": 0.15906637907028198, + "kept_token_count": 655.44375, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9931325078010559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006867489032447338, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7154968261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29002990722654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25840, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.19967724233865738, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.010712754027917982, + "grad_norm": 0.08743511140346527, + "kept_token_count": 655.11875, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9926483571529389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007351635117083788, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7684295654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2735260009766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25850, + "valid_token_count": 818.5640625 + }, + { + "actual_token_mask_ratio": 0.19995927363634108, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.012317827343940735, + "grad_norm": 0.07615502178668976, + "kept_token_count": 654.63125, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0156, + "mgd_cos_sim": 0.9933804512023926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006619550194591284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5511993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5654754638672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25860, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.20064809769392014, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.012115162052214145, + "grad_norm": 0.11186979711055756, + "kept_token_count": 654.3984375, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0157, + "mgd_cos_sim": 0.9928357720375061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007164231594651937, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7554443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98602294921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25870, + "valid_token_count": 818.6640625 + }, + { + "actual_token_mask_ratio": 0.199545282125473, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.010551701113581658, + "grad_norm": 0.16280901432037354, + "kept_token_count": 654.9890625, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9929881870746613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007011815439909697, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1238067626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7301879882813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25880, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.20003905296325683, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.013052381249144674, + "grad_norm": 0.11027394980192184, + "kept_token_count": 654.3296875, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9934283077716828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00657168235629797, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6477386474609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6106292724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25890, + "valid_token_count": 817.95 + }, + { + "actual_token_mask_ratio": 0.2004249155521393, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.012951675290241837, + "grad_norm": 0.12205598503351212, + "kept_token_count": 654.1671875, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0163, + "mgd_cos_sim": 0.9932779848575592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067220116965472695, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2336853027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0876770019531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25900, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.19997246712446212, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.010835571773350238, + "grad_norm": 0.08428949862718582, + "kept_token_count": 654.5015625, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9933479309082032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006652068346738815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9157684326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9647277832031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25910, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.19988655745983125, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.010432391287758946, + "grad_norm": 0.12913985550403595, + "kept_token_count": 654.9328125, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9933330237865448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006666966900229454, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5780487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48427429199216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25920, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.19988936930894852, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.010769044980406761, + "grad_norm": 0.08790285140275955, + "kept_token_count": 654.4546875, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9935442447662354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006455746851861477, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.49176025390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.38364868164064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25930, + "valid_token_count": 817.959375 + }, + { + "actual_token_mask_ratio": 0.19931027740240098, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.01147298295982182, + "grad_norm": 0.09782161563634872, + "kept_token_count": 655.0078125, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9933666586875916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006633343640714884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.54722900390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3003387451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25940, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.20060149282217027, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.01115205567330122, + "grad_norm": 0.09626442939043045, + "kept_token_count": 654.10625, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.99331716299057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006682842411100865, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4626220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1290649414062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25950, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20032214671373366, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.010835159197449684, + "grad_norm": 0.12273886054754257, + "kept_token_count": 654.6078125, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934215903282165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006578413303941488, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6961700439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.13372192382815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25960, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.199512842297554, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.011236036010086537, + "grad_norm": 0.08933927863836288, + "kept_token_count": 655.2015625, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9934013426303864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006598650012165308, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0813842773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.19945068359374, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25970, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.1999750480055809, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.013556085107848049, + "grad_norm": 0.08490067720413208, + "kept_token_count": 654.784375, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9934118449687958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588164437562227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1201995849609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5932373046875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25980, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2003620982170105, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.01293677007779479, + "grad_norm": 0.08421209454536438, + "kept_token_count": 654.5078125, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0164, + "mgd_cos_sim": 0.993012273311615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006987735722213983, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.460302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2134979248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 25990, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.19971020370721818, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.010985025949776173, + "grad_norm": 0.09508924931287766, + "kept_token_count": 654.6765625, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9928962051868438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007103804033249617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.5449615478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.16521606445315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26000, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.19924879372119902, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.012952590640634299, + "grad_norm": 0.12282763421535492, + "kept_token_count": 655.165625, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0168, + "mgd_cos_sim": 0.9923990964889526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007600909005850554, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.7016571044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0147003173828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26010, + "valid_token_count": 818.190625 + }, + { + "actual_token_mask_ratio": 0.20043669492006302, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.011567182000726461, + "grad_norm": 0.1672508716583252, + "kept_token_count": 654.20625, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9930513203144073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069486782886087894, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.9632598876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1570343017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26020, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.1989391028881073, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.010930914618074893, + "grad_norm": 0.12090902030467987, + "kept_token_count": 655.84375, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9926505148410797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007349493633955717, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.90917053222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.72699584960935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26030, + "valid_token_count": 818.7265625 + }, + { + "actual_token_mask_ratio": 0.19995877593755723, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.0126322234980762, + "grad_norm": 0.09653322398662567, + "kept_token_count": 654.33125, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9932810723781585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006718928273767233, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.6909881591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.60289306640624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26040, + "valid_token_count": 817.8765625 + }, + { + "actual_token_mask_ratio": 0.1997199222445488, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.010618301387876273, + "grad_norm": 0.08583679795265198, + "kept_token_count": 654.8, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.014, + "mgd_cos_sim": 0.993163400888443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006836597714573145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1335388183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.94525146484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26050, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.19965920299291612, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.01323352549225092, + "grad_norm": 0.09949702769517899, + "kept_token_count": 654.7984375, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0165, + "mgd_cos_sim": 0.9935017883777618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006498222518712282, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8894836425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0024810791016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26060, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.199899822473526, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.010438406467437744, + "grad_norm": 0.09113085269927979, + "kept_token_count": 654.965625, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9937146723270416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006285321433097124, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4940124511719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5521759033203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26070, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.20004488080739974, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.01101325531490147, + "grad_norm": 0.07179344445466995, + "kept_token_count": 654.253125, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934633433818817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006536676269024611, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.61663513183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.700927734375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26080, + "valid_token_count": 817.8640625 + }, + { + "actual_token_mask_ratio": 0.20007449239492417, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.009453028393909334, + "grad_norm": 0.10883410274982452, + "kept_token_count": 654.6203125, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9932453095912933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006754678208380938, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.49107666015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3461212158203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26090, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.1995732828974724, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.01106315297074616, + "grad_norm": 0.09489606320858002, + "kept_token_count": 655.1578125, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933316290378571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006668362300843, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1423278808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.61619262695314, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26100, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.2017694890499115, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.013478219555690885, + "grad_norm": 0.08143826574087143, + "kept_token_count": 653.2265625, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0169, + "mgd_cos_sim": 0.9932527661323547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747233774513006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4756622314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3938781738281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26110, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.20006378442049028, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.01279422789812088, + "grad_norm": 0.1235823854804039, + "kept_token_count": 654.525, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0162, + "mgd_cos_sim": 0.993113887310028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068861118517816065, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7193328857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2626647949219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26120, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20077210217714309, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.010964453918859363, + "grad_norm": 0.1048709899187088, + "kept_token_count": 653.865625, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9930975556373596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006902451533824206, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7817718505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4458892822266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26130, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.19889241307973862, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.011367191374301911, + "grad_norm": 0.11345934122800827, + "kept_token_count": 655.446875, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9931143939495086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006885604653507471, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.42954711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.5430419921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2008809819817543, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.011663139471784234, + "grad_norm": 0.10246934741735458, + "kept_token_count": 653.8765625, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9929229855537415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007077024225145579, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.80162353515624, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26150, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.20054641962051392, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.011225581681355835, + "grad_norm": 0.09628476202487946, + "kept_token_count": 653.871875, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9926629841327668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073370193131268024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.72296752929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1872863769531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26160, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.1995876595377922, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.0115085294470191, + "grad_norm": 0.10194623470306396, + "kept_token_count": 655.2484375, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9935308933258057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006469113472849131, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.06670532226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5899627685547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26170, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.1997399091720581, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.010270506143569946, + "grad_norm": 0.08493208140134811, + "kept_token_count": 654.5953125, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9931785583496093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068214344792068005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.0619812011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.60411071777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26180, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.20005762726068496, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.011258467007428408, + "grad_norm": 0.12090262770652771, + "kept_token_count": 654.1828125, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9929587662220001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007041233032941819, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9867919921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7355987548828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26190, + "valid_token_count": 817.7875 + }, + { + "actual_token_mask_ratio": 0.19952171593904494, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.011414336087182165, + "grad_norm": 0.1560690850019455, + "kept_token_count": 655.1953125, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9930588066577911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941199209541083, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3193420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.57518005371094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26200, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.2004635825753212, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.01037209746427834, + "grad_norm": 0.0979096069931984, + "kept_token_count": 654.3265625, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9932745337486267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725478079169988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9015777587891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.92059936523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26210, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20006371587514876, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.011872747726738454, + "grad_norm": 0.10017595440149307, + "kept_token_count": 654.6484375, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9928008615970612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007199152279645205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8763397216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15887756347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26220, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.20034174621105194, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.011829842440783977, + "grad_norm": 0.07999015599489212, + "kept_token_count": 654.28125, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9931608736515045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006839143298566342, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8116027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2532440185547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26230, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2004626661539078, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.010976124368607999, + "grad_norm": 0.10368889570236206, + "kept_token_count": 654.0390625, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9936209261417389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006379061657935381, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.14362182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5674621582031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26240, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.2004730224609375, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.009833304770290851, + "grad_norm": 0.09791561216115952, + "kept_token_count": 654.25625, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9934651494026184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065348540432751175, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1983276367188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2821472167969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26250, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.20025506168603896, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.014213920012116433, + "grad_norm": 0.07735208421945572, + "kept_token_count": 654.4, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9933508396148681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00664915619418025, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.69732055664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8441497802734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26260, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.20068505257368088, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.010142941446974873, + "grad_norm": 0.07893894612789154, + "kept_token_count": 654.1078125, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9931889832019806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006811005715280771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.18475341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.20456848144534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.19959839284420014, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.011905877804383635, + "grad_norm": 0.08589693158864975, + "kept_token_count": 655.0109375, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9928978025913239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007102202344685793, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.93580932617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.66762084960936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26280, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20021393448114394, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.010819543758407235, + "grad_norm": 0.09775470197200775, + "kept_token_count": 654.084375, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9934871912002563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065128143876791, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9101989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.42491149902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26290, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.2000106766819954, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.00952491988427937, + "grad_norm": 0.08643550425767899, + "kept_token_count": 655.0046875, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9932154715061188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006784532964229584, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9220031738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5790496826172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26300, + "valid_token_count": 818.771875 + }, + { + "actual_token_mask_ratio": 0.19987194687128068, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.014001141581684352, + "grad_norm": 0.09777262061834335, + "kept_token_count": 654.9796875, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0172, + "mgd_cos_sim": 0.9936438858509063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006356105301529169, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6890930175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2209899902344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26310, + "valid_token_count": 818.5953125 + }, + { + "actual_token_mask_ratio": 0.1992191568017006, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.010455051017925144, + "grad_norm": 0.10315403342247009, + "kept_token_count": 655.253125, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9928570091724396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007142992038279772, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5544097900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7687194824219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26320, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.20008160769939423, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.012298908922821284, + "grad_norm": 0.05715400353074074, + "kept_token_count": 655.03125, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9936587274074554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341259181499481, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3415008544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0208801269531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26330, + "valid_token_count": 818.8765625 + }, + { + "actual_token_mask_ratio": 0.1995449185371399, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.01513985712081194, + "grad_norm": 0.10457314550876617, + "kept_token_count": 655.015625, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0185, + "mgd_cos_sim": 0.9932191550731659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006780857220292091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8684783935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6022644042969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26340, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.20127061754465103, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.01123072816990316, + "grad_norm": 0.10092084109783173, + "kept_token_count": 653.35, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933956921100616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066043084487318994, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0137664794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7318054199219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26350, + "valid_token_count": 817.984375 + }, + { + "actual_token_mask_ratio": 0.19977409094572068, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.009429731033742427, + "grad_norm": 0.0971309095621109, + "kept_token_count": 654.6921875, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9934037446975708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065962646156549455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.46900329589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7512969970703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26360, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.200577612221241, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.011674954509362578, + "grad_norm": 0.10333124548196793, + "kept_token_count": 654.24375, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9935792624950409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006420743279159069, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1800842285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4252502441406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26370, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.20037686824798584, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.012580929649993777, + "grad_norm": 0.12239215523004532, + "kept_token_count": 654.48125, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0161, + "mgd_cos_sim": 0.9929485023021698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007051485031843186, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2835418701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8253662109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26380, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.1994885042309761, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.013346746377646924, + "grad_norm": 0.10998354107141495, + "kept_token_count": 654.7703125, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0166, + "mgd_cos_sim": 0.9934534907341004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006546510290354491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9799011230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.1471374511719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26390, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20035184025764466, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.012519181426614522, + "grad_norm": 0.09632965922355652, + "kept_token_count": 654.4609375, + "learning_rate": 1.166292824099105e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9931226670742035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006877330597490072, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.55654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.89470825195315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26400, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2002308800816536, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.009521323768422008, + "grad_norm": 0.08520200848579407, + "kept_token_count": 654.6796875, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9933946073055268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006605401355773211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.4391143798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5889099121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26410, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.20030723959207536, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.010332764592021704, + "grad_norm": 0.09779080003499985, + "kept_token_count": 654.365625, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9932170033454895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006782986875623465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.48825988769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3393615722656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26420, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2000397190451622, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.012269401457160711, + "grad_norm": 0.09825160354375839, + "kept_token_count": 654.484375, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0158, + "mgd_cos_sim": 0.9930363178253174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963698565959931, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0027221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3573486328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26430, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.20047499388456344, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.0113779719453305, + "grad_norm": 0.09022367745637894, + "kept_token_count": 654.425, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936670124530792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006332988198846578, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.63310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8489685058594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26440, + "valid_token_count": 818.515625 + }, + { + "actual_token_mask_ratio": 0.19916584640741347, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.009948401246219873, + "grad_norm": 0.10019976645708084, + "kept_token_count": 655.353125, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938738942146301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006126104015856981, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.5212066650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8361511230469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26450, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2002347618341446, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.010084359720349313, + "grad_norm": 0.07463739067316055, + "kept_token_count": 654.48125, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9933772385120392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066227615810930725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2430908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5261627197266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26460, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.19952517300844191, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.011863712687045336, + "grad_norm": 0.11881095916032791, + "kept_token_count": 654.9859375, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9931469976902008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006853010039776564, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.91484985351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98585510253906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26470, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.19864056408405303, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.011885617207735778, + "grad_norm": 0.10735098272562027, + "kept_token_count": 655.6140625, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9933374285697937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006662577204406261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1011627197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62303161621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26480, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2008395791053772, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.011106753628700972, + "grad_norm": 0.11699353903532028, + "kept_token_count": 654.0078125, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933960437774658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006603953335434199, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.92104187011716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9731384277344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26490, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.19959123730659484, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.011508733872324229, + "grad_norm": 0.09095790982246399, + "kept_token_count": 654.8, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9932624101638794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006737608648836613, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0861480712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48109741210936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26500, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.20043291449546813, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.012072557350620627, + "grad_norm": 0.08799639344215393, + "kept_token_count": 654.365625, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0153, + "mgd_cos_sim": 0.9935793995857238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00642058439552784, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5572998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5062286376953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26510, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2005437806248665, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.009796064207330346, + "grad_norm": 0.10373137891292572, + "kept_token_count": 654.39375, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9938035607337952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006196431908756494, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.7369079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.09270935058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26520, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.19960891753435134, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.012410191632807255, + "grad_norm": 0.08205756545066833, + "kept_token_count": 654.690625, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.993752408027649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247591320425272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.24140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0839782714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26530, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.19899183213710786, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.01083943471312523, + "grad_norm": 0.08386211842298508, + "kept_token_count": 655.303125, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9930124938488006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006987522635608911, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.54798278808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4507354736328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26540, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.19984472095966338, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.011386883119121194, + "grad_norm": 0.10695866495370865, + "kept_token_count": 654.8, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.993468290567398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006531716696918011, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.28893127441404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.55914306640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26550, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.20079777687788009, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.010975290276110172, + "grad_norm": 0.10715439915657043, + "kept_token_count": 653.9828125, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9931488454341888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068511394783854485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9091430664063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.08079223632814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26560, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.20011248141527177, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.01531803011894226, + "grad_norm": 0.08737888187170029, + "kept_token_count": 654.45625, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0185, + "mgd_cos_sim": 0.9935854077339172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006414597574621439, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8619445800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.09546813964846, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26570, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20072656720876694, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.011795490747317672, + "grad_norm": 0.10368400067090988, + "kept_token_count": 653.978125, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.015, + "mgd_cos_sim": 0.993656599521637, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006343394145369529, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7240478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2526458740234, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26580, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.1990688368678093, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.010162550816312432, + "grad_norm": 0.07007072120904922, + "kept_token_count": 655.296875, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.993497657775879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006502342969179153, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7482666015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.8848937988281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26590, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.19968140125274658, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.013850918086245657, + "grad_norm": 0.11690040677785873, + "kept_token_count": 654.8015625, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0175, + "mgd_cos_sim": 0.9926792085170746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007320801913738251, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3629455566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.49968872070315, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26600, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.20031943768262864, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.011589155578985811, + "grad_norm": 0.0970916748046875, + "kept_token_count": 654.2921875, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9934181332588196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006581865437328816, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1304046630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.94774169921874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26610, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.20017921328544616, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.009880079515278339, + "grad_norm": 0.11622271686792374, + "kept_token_count": 654.428125, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9931690216064453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006830960232764483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.25291137695314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1162139892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26620, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19999316483736038, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.012156393053010106, + "grad_norm": 0.09260367602109909, + "kept_token_count": 654.5890625, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0155, + "mgd_cos_sim": 0.9933629691600799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006637024786323309, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.69378356933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.67255859375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26630, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2009537249803543, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.012074028747156263, + "grad_norm": 0.06771053373813629, + "kept_token_count": 654.0078125, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936727941036224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006327198725193739, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.15528869628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5371856689453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26640, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.20062404423952102, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.010878712125122546, + "grad_norm": 0.11758501082658768, + "kept_token_count": 654.1203125, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9930969655513764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006903035566210747, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.8580322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6466827392578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26650, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.19940390884876252, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.01108094616793096, + "grad_norm": 0.06616683304309845, + "kept_token_count": 655.040625, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9934074699878692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006592541746795177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.28358459472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9164978027344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26660, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.20094113945960998, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.011005544522777199, + "grad_norm": 0.0707639530301094, + "kept_token_count": 653.8796875, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9932143688201904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067856271751225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9519989013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0094360351562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26670, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19990097731351852, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.010809809155762195, + "grad_norm": 0.09982557594776154, + "kept_token_count": 654.6984375, + "learning_rate": 9.939247615320047e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9935499966144562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006450010370463133, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.0188751220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0038299560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26680, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.19978934973478318, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.01124001843854785, + "grad_norm": 0.09062503278255463, + "kept_token_count": 654.9390625, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9933617055416107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006638316344469786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.74452514648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2499725341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26690, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.19964322447776794, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.012172393407672644, + "grad_norm": 0.08608540147542953, + "kept_token_count": 655.034375, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9935116291046142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064883572980761525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9017333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7678741455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26700, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.19935871213674544, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.011834646807983518, + "grad_norm": 0.08354484289884567, + "kept_token_count": 655.3140625, + "learning_rate": 9.762452810644068e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9936381161212922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006361872982233762, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2778259277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5493499755859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26710, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.20032705664634703, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.012339633097872138, + "grad_norm": 0.08758294582366943, + "kept_token_count": 654.209375, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0158, + "mgd_cos_sim": 0.9929949045181274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007005106192082167, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2183471679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8358520507812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26720, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.19960920363664628, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.01501933466643095, + "grad_norm": 0.11389076709747314, + "kept_token_count": 654.8921875, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0184, + "mgd_cos_sim": 0.9931629598140717, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006837050803005696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.98643493652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.78697814941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26730, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2002342849969864, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.011553947906941175, + "grad_norm": 0.0816541463136673, + "kept_token_count": 654.2, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9927775144577027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007222490385174751, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.77393493652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.43629760742186, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26740, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.20084174424409867, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.010074327513575555, + "grad_norm": 0.06958536803722382, + "kept_token_count": 654.1078125, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9933357834815979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006664191000163555, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4188690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.42604064941406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26750, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.19936760514974594, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.012590848561376333, + "grad_norm": 0.09251399338245392, + "kept_token_count": 655.2890625, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9932870686054229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00671292170882225, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8502532958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5362579345703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26760, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.20055821985006334, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.011560972826555372, + "grad_norm": 0.06176471337676048, + "kept_token_count": 654.0421875, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9937530636787415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062469481490552425, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1097839355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0418640136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26770, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.19979450553655625, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.010678933188319206, + "grad_norm": 0.06866131722927094, + "kept_token_count": 654.9515625, + "learning_rate": 9.355898969692889e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9933869481086731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613059341907502, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5553833007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.027099609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26780, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.20057126581668855, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.013140468299388886, + "grad_norm": 0.10635197907686234, + "kept_token_count": 654.190625, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0165, + "mgd_cos_sim": 0.9932791233062744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006720876693725586, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.44920959472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0954864501953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26790, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2004105880856514, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.011874099960550666, + "grad_norm": 0.1074914038181305, + "kept_token_count": 654.2453125, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.993319857120514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006680138781666756, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.31089477539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2047515869141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26800, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.1992810770869255, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.010295610502362251, + "grad_norm": 0.10060980170965195, + "kept_token_count": 655.2171875, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9934293925762177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006570624001324177, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.27561645507814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6087982177734, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.19921448826789856, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.010467049153521656, + "grad_norm": 0.11619565635919571, + "kept_token_count": 655.184375, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9930752515792847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006924749445170164, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.25415954589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.86102600097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26820, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.1999944418668747, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.011898099491372705, + "grad_norm": 0.05698242038488388, + "kept_token_count": 654.7828125, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9931667745113373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006833227910101413, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2144287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2204956054687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26830, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.19969662874937058, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.009761211089789867, + "grad_norm": 0.08358188718557358, + "kept_token_count": 654.85625, + "learning_rate": 9.014088250486052e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9935410380363464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006458954233676195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1174285888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5965270996094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26840, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.20022303909063338, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.011999090760946273, + "grad_norm": 0.11148684471845627, + "kept_token_count": 654.49375, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9926552712917328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007344729639589787, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7756591796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0725494384766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26850, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19951872825622557, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.011699270363897086, + "grad_norm": 0.10971521586179733, + "kept_token_count": 654.975, + "learning_rate": 8.90152109804091e-07, + "loss": 0.015, + "mgd_cos_sim": 0.993470299243927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006529708858579397, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9458068847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.08252868652346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26860, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20069230645895003, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.010712066572159529, + "grad_norm": 0.08597913384437561, + "kept_token_count": 653.9765625, + "learning_rate": 8.845494659610365e-07, + "loss": 0.014, + "mgd_cos_sim": 0.993380606174469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006619388330727815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.3890106201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2300720214844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26870, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2002352148294449, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.010665546031668782, + "grad_norm": 0.10171252489089966, + "kept_token_count": 654.6625, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9936875224113464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006312486249953508, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2690002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.41294555664064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26880, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.1986330345273018, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.010511660249903798, + "grad_norm": 0.07419517636299133, + "kept_token_count": 655.609375, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9934543609619141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545631028711796, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4521759033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.65216369628905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26890, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.2009831964969635, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.010746457055211068, + "grad_norm": 0.06618090718984604, + "kept_token_count": 653.8125, + "learning_rate": 8.678444711790157e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9934705913066864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065294202417135235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.89520263671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6200653076172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26900, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20049123615026473, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.01051446287892759, + "grad_norm": 0.10272025316953659, + "kept_token_count": 654.2734375, + "learning_rate": 8.623104743875493e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9929891705513001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007010814361274242, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5021667480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.67864685058595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26910, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20073701739311217, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.009903769800439477, + "grad_norm": 0.11072326451539993, + "kept_token_count": 653.9734375, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9934259235858918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006574062723666429, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7010498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5141204833984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26920, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2008462980389595, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.011527342535555363, + "grad_norm": 0.09206406027078629, + "kept_token_count": 653.884375, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9935691773891449, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006430826988071204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3866424560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.70025024414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26930, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20040978640317916, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.009263546857982873, + "grad_norm": 0.07620784640312195, + "kept_token_count": 654.296875, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9933179080486297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006682084035128355, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8356201171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.16650390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26940, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.20018249601125718, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.011323458747938275, + "grad_norm": 0.08256898820400238, + "kept_token_count": 654.41875, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9938061773777008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006193821411579847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8447784423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4027801513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26950, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.1992135837674141, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.011894702073186636, + "grad_norm": 0.14168578386306763, + "kept_token_count": 655.171875, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9932672441005707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006732761114835739, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9164093017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.96513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26960, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2002093732357025, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.008769348543137313, + "grad_norm": 0.13457557559013367, + "kept_token_count": 654.2921875, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9939061820507049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006093833409249782, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6489715576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.02626037597656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26970, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19992138296365738, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.011860205559059978, + "grad_norm": 0.08903463184833527, + "kept_token_count": 654.71875, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9933310210704803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006668964121490717, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.81116943359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.68016052246094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26980, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.1999051481485367, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.01141859358176589, + "grad_norm": 0.09644002467393875, + "kept_token_count": 654.5515625, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9931524991989136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006847506947815418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.15289611816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10601196289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 26990, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.1996842607855797, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.011862338427454234, + "grad_norm": 0.10269796848297119, + "kept_token_count": 655.0015625, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.993151742219925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068482448346912864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.06542053222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3865600585938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27000, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.19976240992546082, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.010438572894781829, + "grad_norm": 0.08699806779623032, + "kept_token_count": 654.540625, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9939886569976807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006011347100138664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.5324249267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.007666015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27010, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.19918008148670197, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.011503913206979632, + "grad_norm": 0.06838672608137131, + "kept_token_count": 655.275, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9936703383922577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006329656112939119, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9581695556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0802886962891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27020, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.19934101551771163, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.011882568476721644, + "grad_norm": 0.08508841693401337, + "kept_token_count": 655.2265625, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9935814797878265, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418516300618649, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.6277221679687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9313232421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27030, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19972661584615709, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.011997421132400632, + "grad_norm": 0.0865139588713646, + "kept_token_count": 655.071875, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9933227837085724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006677223090082407, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.71885986328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.73882446289065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27040, + "valid_token_count": 818.5625 + }, + { + "actual_token_mask_ratio": 0.20061751902103425, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.011497169500216842, + "grad_norm": 0.10245434194803238, + "kept_token_count": 654.3671875, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9933316886425019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066683161072433, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1689819335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6487152099609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27050, + "valid_token_count": 818.590625 + }, + { + "actual_token_mask_ratio": 0.19984029233455658, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.011441676504909992, + "grad_norm": 0.0939566120505333, + "kept_token_count": 655.0109375, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9934983611106872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006501623895019293, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2555908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.85431213378905, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27060, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.20022416859865189, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.013385079428553581, + "grad_norm": 0.11990451067686081, + "kept_token_count": 654.4203125, + "learning_rate": 7.761073274428076e-07, + "loss": 0.017, + "mgd_cos_sim": 0.9927217245101929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072782646864652635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.07347106933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1424194335938, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27070, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20060216039419174, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.013319128472357989, + "grad_norm": 0.09339453279972076, + "kept_token_count": 654.153125, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0166, + "mgd_cos_sim": 0.9933822453022003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617749109864235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.3119689941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3841186523438, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27080, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.20039083063602448, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.01010135654360056, + "grad_norm": 0.09710875898599625, + "kept_token_count": 654.3109375, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.99345263838768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065473650582134725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.80503845214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.62245178222656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.19948440045118332, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.012183861341327429, + "grad_norm": 0.09861625730991364, + "kept_token_count": 654.88125, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9930742025375366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069257961586117744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7705810546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.94806518554685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27100, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19972600936889648, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.01041608229279518, + "grad_norm": 0.10287362337112427, + "kept_token_count": 654.8875, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9934813380241394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006518675200641155, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.83412475585936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.75575561523436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27110, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.19937360733747483, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.011134784761816263, + "grad_norm": 0.06598047912120819, + "kept_token_count": 655.1171875, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9935610890388489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006438911240547895, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3472076416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9257049560547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27120, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.19953869879245759, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.011956616165116429, + "grad_norm": 0.11650747060775757, + "kept_token_count": 655.340625, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9936624646186829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006337528862059116, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.5510986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1158447265625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27130, + "valid_token_count": 818.7015625 + }, + { + "actual_token_mask_ratio": 0.19950173199176788, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.0109906867146492, + "grad_norm": 0.09814997017383575, + "kept_token_count": 655.0578125, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9931509137153626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006849084049463272, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.61724548339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7751983642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27140, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.20108221024274825, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.010084498021751643, + "grad_norm": 0.08391432464122772, + "kept_token_count": 653.5734375, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9933950424194335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006604958325624466, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3956665039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6438018798828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27150, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.19994455724954605, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.009343784162774682, + "grad_norm": 0.09458832442760468, + "kept_token_count": 654.75, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9934789121150971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065210903994739056, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3374359130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.12265930175784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27160, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20026568621397017, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.010535871284082532, + "grad_norm": 0.07511664181947708, + "kept_token_count": 654.575, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.993727833032608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00627214414998889, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3336181640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3225372314453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27170, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.20002013444900513, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.013300239248201251, + "grad_norm": 0.14721891283988953, + "kept_token_count": 654.4703125, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0167, + "mgd_cos_sim": 0.993134742975235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068652602843940255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.53057250976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7341735839844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27180, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.20000426471233368, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.011257100477814674, + "grad_norm": 0.09523504227399826, + "kept_token_count": 654.721875, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9932068228721619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006793183088302612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8625122070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0924926757813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27190, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.2004848450422287, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.011795573029667139, + "grad_norm": 0.08436950296163559, + "kept_token_count": 654.3828125, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9931485414505005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006851454172283411, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.7199401855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9686981201172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27200, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.19946640133857726, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.012031714897602797, + "grad_norm": 0.09900856018066406, + "kept_token_count": 655.225, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9931828200817108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068171733058989045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.0362884521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4134033203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27210, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.19981739521026612, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.011037223227322101, + "grad_norm": 0.10542377084493637, + "kept_token_count": 654.7578125, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9934858024120331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006514204945415258, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.580224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.79547119140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27220, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19900898039340972, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.010257774172350765, + "grad_norm": 0.10601446777582169, + "kept_token_count": 655.5015625, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9933489978313446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006651014089584351, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.0377258300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0764526367187, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27230, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19945442974567412, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.011005586944520473, + "grad_norm": 0.09743732959032059, + "kept_token_count": 655.04375, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9935737490653992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064262707717716696, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.13087768554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.12945861816405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27240, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.20002396702766417, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.01188124604523182, + "grad_norm": 0.09424231201410294, + "kept_token_count": 654.5359375, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9936432123184205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006356790289282798, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.82911071777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.19935620576143265, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.01271052546799183, + "grad_norm": 0.09118661284446716, + "kept_token_count": 655.096875, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0161, + "mgd_cos_sim": 0.9931880891323089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006811894476413727, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.42205505371095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.16583251953125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27260, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.19966121762990952, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.010076931957155467, + "grad_norm": 0.1449003964662552, + "kept_token_count": 654.8515625, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9936566889286041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006343307159841061, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.0032653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.77448120117185, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27270, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.1998743534088135, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.011337010841816663, + "grad_norm": 0.0829632356762886, + "kept_token_count": 654.80625, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9932653725147247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006734616868197918, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.25242614746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.28460083007815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27280, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.20088555961847304, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.010746151814237237, + "grad_norm": 0.11104234308004379, + "kept_token_count": 653.6265625, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9933144330978394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00668555898591876, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8769165039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.50978088378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27290, + "valid_token_count": 817.9421875 + }, + { + "actual_token_mask_ratio": 0.20037085562944412, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.011038914928212762, + "grad_norm": 0.08615925908088684, + "kept_token_count": 654.2859375, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9933554053306579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006644598487764597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.4175201416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0849914550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27300, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.1990702345967293, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.011032738583162427, + "grad_norm": 0.07056289166212082, + "kept_token_count": 655.4515625, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.9935516536235809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448338925838471, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.4434051513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4703125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27310, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.19993814677000046, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.011356421466916799, + "grad_norm": 0.10157383978366852, + "kept_token_count": 654.8015625, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9938504040241242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006149594951421022, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5002105712891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.19078674316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27320, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.19992799311876297, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.010899536777287722, + "grad_norm": 0.0981406718492508, + "kept_token_count": 654.4953125, + "learning_rate": 6.454750509527874e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9937678873538971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006232117302715778, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.88291931152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4240356445313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27330, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.2000936061143875, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.011613320652395486, + "grad_norm": 0.059319477528333664, + "kept_token_count": 654.503125, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9929061591625213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007093836925923825, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9151184082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9962219238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27340, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.20004838556051255, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.01138389278203249, + "grad_norm": 0.09004215896129608, + "kept_token_count": 654.5359375, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9935082077980042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064917923882603645, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8325927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15138854980466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27350, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.199357008934021, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.010536876413971186, + "grad_norm": 0.08249132335186005, + "kept_token_count": 655.3640625, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9934094905853271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006590501498430967, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6127868652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.54355773925784, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27360, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.199657142162323, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.009763649944216014, + "grad_norm": 0.07060381770133972, + "kept_token_count": 654.890625, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9934121072292328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065878950990736485, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7339111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1996276855469, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27370, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20016689449548722, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.01077013174071908, + "grad_norm": 0.10216755419969559, + "kept_token_count": 654.340625, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9930974185466767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00690258014947176, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9064086914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.91424865722655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27380, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.19978861808776854, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.009268867038190364, + "grad_norm": 0.059286169707775116, + "kept_token_count": 654.690625, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9931360721588135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006863916385918855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.89422607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0918426513672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27390, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2008271798491478, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.011398330703377724, + "grad_norm": 0.10557185113430023, + "kept_token_count": 653.9390625, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9934321284294129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006567875761538744, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0025970458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.626708984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.20063070505857467, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.011700380779802799, + "grad_norm": 0.10165541619062424, + "kept_token_count": 653.9859375, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9928690433502197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007130951341241598, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.14812622070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5950164794922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27410, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.20093929320573806, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.009785544639453293, + "grad_norm": 0.09749092161655426, + "kept_token_count": 653.4890625, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9934318602085114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006568146590143442, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.7295257568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7941558837891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27420, + "valid_token_count": 817.821875 + }, + { + "actual_token_mask_ratio": 0.20066432952880858, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.01153965713456273, + "grad_norm": 0.09088177233934402, + "kept_token_count": 653.978125, + "learning_rate": 5.983656853716097e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9931486189365387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006851387955248356, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.98890075683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.32562866210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27430, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.20145600587129592, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.012039671977981925, + "grad_norm": 0.10817170143127441, + "kept_token_count": 653.4015625, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9934990048408509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006500988826155662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.98784790039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.28785400390626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27440, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.1997321590781212, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.012655319506302476, + "grad_norm": 0.09697514772415161, + "kept_token_count": 654.85625, + "learning_rate": 5.89153622838805e-07, + "loss": 0.016, + "mgd_cos_sim": 0.9932413280010224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00675866799429059, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.10821838378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7689483642578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27450, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.20087069571018218, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.012640277668833733, + "grad_norm": 0.10283073782920837, + "kept_token_count": 653.9734375, + "learning_rate": 5.845738545851431e-07, + "loss": 0.016, + "mgd_cos_sim": 0.9932975113391876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00670248344540596, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.30897521972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1552276611328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27460, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2010195031762123, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.00986732537858188, + "grad_norm": 0.06677690893411636, + "kept_token_count": 653.70625, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9934769153594971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006523082312196493, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.7772491455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.722021484375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27470, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.19992077946662903, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.011019839765504002, + "grad_norm": 0.12017382681369781, + "kept_token_count": 654.7984375, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0143, + "mgd_cos_sim": 0.993436074256897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006563913729041815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.88670654296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5377563476562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27480, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.19967055469751357, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.010009045177139343, + "grad_norm": 0.04788250848650932, + "kept_token_count": 654.68125, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9942596614360809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005740337166935205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.55083923339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.44530029296874, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27490, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.20008435249328613, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.009931819234043359, + "grad_norm": 0.07794281095266342, + "kept_token_count": 654.6578125, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9935833930969238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006416631396859884, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.37079162597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4514404296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27500, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.2001499742269516, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.0111541744787246, + "grad_norm": 0.11200428009033203, + "kept_token_count": 654.2328125, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9930369079113006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069630882702767845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0586456298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8065979003906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27510, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.20094882547855378, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.012934977468103171, + "grad_norm": 0.1029329001903534, + "kept_token_count": 653.9375, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0162, + "mgd_cos_sim": 0.9934378266334534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00656217997893691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.29930725097654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1109100341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27520, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.19978010505437852, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010003639245405793, + "grad_norm": 0.11256629228591919, + "kept_token_count": 655.0, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9932558000087738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006744206510484218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6517852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.12973937988284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27530, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19957343935966493, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.011552163865417242, + "grad_norm": 0.08599721640348434, + "kept_token_count": 655.2109375, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9933861076831818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613872665911913, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7297119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7075531005859, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27540, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.20005222111940385, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.010584998177364469, + "grad_norm": 0.06251116096973419, + "kept_token_count": 654.5515625, + "learning_rate": 5.441448256498544e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9932513773441315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006748619955033064, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6010284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8590850830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27550, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.20068893134593963, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.01285620341077447, + "grad_norm": 0.09295962750911713, + "kept_token_count": 654.0078125, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0162, + "mgd_cos_sim": 0.9933535993099213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006646407674998045, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.3664031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6444396972656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27560, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.19906865656375886, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.011397707974538207, + "grad_norm": 0.0958242118358612, + "kept_token_count": 655.6125, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9938079059123993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192097160965204, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9776153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6383911132813, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27570, + "valid_token_count": 818.565625 + }, + { + "actual_token_mask_ratio": 0.20067329108715057, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.009024051064625383, + "grad_norm": 0.08858916163444519, + "kept_token_count": 654.246875, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.9934432327747345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006556772813200951, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.76483764648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3941131591797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27580, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.20064650177955629, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.010071322973817586, + "grad_norm": 0.06246722862124443, + "kept_token_count": 653.96875, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9932572484016419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006742747314274311, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.78615417480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7596801757812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27590, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.20054250061511994, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.011370568815618754, + "grad_norm": 0.10245709866285324, + "kept_token_count": 654.3734375, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9934658825397491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006534113362431526, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.3783325195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1296081542969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27600, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.19960042387247084, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.00950173861347139, + "grad_norm": 0.08565723896026611, + "kept_token_count": 654.8859375, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9935958623886109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006404166575521231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0944580078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1412078857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27610, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2002193436026573, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.012132055079564451, + "grad_norm": 0.09426003694534302, + "kept_token_count": 654.709375, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9935480535030365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006451945938169956, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.18128356933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.34766845703126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27620, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.20015748739242553, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.011991084180772304, + "grad_norm": 0.14363142848014832, + "kept_token_count": 654.6609375, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9928958535194397, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007104130461812019, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2704620361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7308349609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27630, + "valid_token_count": 818.484375 + }, + { + "actual_token_mask_ratio": 0.19939377903938293, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.012781191943213343, + "grad_norm": 0.08798251301050186, + "kept_token_count": 655.190625, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0161, + "mgd_cos_sim": 0.9933550357818604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006644968595355749, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.36647338867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.18846740722654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27640, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.1988646298646927, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.011085157003253698, + "grad_norm": 0.09603812545537949, + "kept_token_count": 655.521875, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.99334157705307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006658430770039558, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0312561035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7539031982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27650, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.1994759738445282, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.011366199236363173, + "grad_norm": 0.09719634801149368, + "kept_token_count": 655.25625, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9930881142616272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00691186711192131, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.0072418212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1431030273437, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27660, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.1999365732073784, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.010571133531630039, + "grad_norm": 0.09666763991117477, + "kept_token_count": 654.575, + "learning_rate": 4.924533220104521e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9932366490364075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006763349380344152, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8549041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5913879394531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27670, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.19893214404582976, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.011402562772855163, + "grad_norm": 0.09343959391117096, + "kept_token_count": 655.325, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0146, + "mgd_cos_sim": 0.9936179161071778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006382081378251314, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.89686279296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.353369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27680, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.198902890086174, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.013636492751538753, + "grad_norm": 0.08267321437597275, + "kept_token_count": 655.65, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0171, + "mgd_cos_sim": 0.9931016385555267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898359023034572, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1706115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.54705505371095, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27690, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2001286044716835, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.009647687105461955, + "grad_norm": 0.06631548702716827, + "kept_token_count": 654.1890625, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9935088276863098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006491183023899794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2359558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1551910400391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27700, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.19987863898277283, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.012309466442093252, + "grad_norm": 0.10206803679466248, + "kept_token_count": 654.4828125, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9931497991085052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006850217841565609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.03583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0357635498047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27710, + "valid_token_count": 817.978125 + }, + { + "actual_token_mask_ratio": 0.19994044303894043, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.014688170794397593, + "grad_norm": 0.10233897715806961, + "kept_token_count": 654.70625, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0181, + "mgd_cos_sim": 0.993234270811081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006765729002654553, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.2240020751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6983703613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27720, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.19956953972578048, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.011726007098332047, + "grad_norm": 0.1017778292298317, + "kept_token_count": 654.740625, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.993217259645462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006782732903957367, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.53418884277346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 385.8592834472656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27730, + "valid_token_count": 817.98125 + }, + { + "actual_token_mask_ratio": 0.19948955178260802, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.01080600805580616, + "grad_norm": 0.08768416941165924, + "kept_token_count": 654.9046875, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.993446284532547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006553720030933618, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.92515258789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8551025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27740, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.20044082552194595, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.011945712100714444, + "grad_norm": 0.07867632061243057, + "kept_token_count": 654.1109375, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9932133853435516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006786618381738663, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6753814697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4218292236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27750, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.199350069463253, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.011401726072654128, + "grad_norm": 0.11089270561933517, + "kept_token_count": 655.259375, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9932697296142579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006730262376368046, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1073028564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.71807556152345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27760, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.19958851486444473, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.009543411107733846, + "grad_norm": 0.08350730687379837, + "kept_token_count": 655.0125, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9933158040046692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006684184819459915, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.0062194824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6546905517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27770, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.20025818943977355, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.011673183133825659, + "grad_norm": 0.10442149639129639, + "kept_token_count": 654.3875, + "learning_rate": 4.472991452817099e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9934151113033295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006584879662841559, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.51868896484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.47109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27780, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19945078343153, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.011786401877179742, + "grad_norm": 0.0826365202665329, + "kept_token_count": 655.1859375, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9931467115879059, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006853283382952213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.77935180664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.28284912109376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27790, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.2006799876689911, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.012367214960977434, + "grad_norm": 0.12453852593898773, + "kept_token_count": 653.9796875, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0156, + "mgd_cos_sim": 0.9935491502285003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006450869236141443, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.55016174316404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2943389892578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27800, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.19928230047225953, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.01141214738599956, + "grad_norm": 0.09681009501218796, + "kept_token_count": 655.0890625, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9930753350257874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006924663577228785, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.53562316894534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.3784576416016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27810, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.1991359382867813, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.01269122464582324, + "grad_norm": 0.07468855381011963, + "kept_token_count": 655.2953125, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9934942960739136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006505694519728422, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.94651489257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23916015625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27820, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.20053063631057738, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.012133245822042227, + "grad_norm": 0.08785043656826019, + "kept_token_count": 654.15625, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0158, + "mgd_cos_sim": 0.9927585482597351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072414446622133255, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.27911071777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.98008422851564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27830, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2005653962492943, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.00952699352055788, + "grad_norm": 0.15251685678958893, + "kept_token_count": 654.0625, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9924519240856171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007548067532479763, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0662780761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0820007324219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27840, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.19989087283611298, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.010681225126609207, + "grad_norm": 0.06012355908751488, + "kept_token_count": 654.778125, + "learning_rate": 4.196782726807097e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9933323383331298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00666766706854105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1387451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7895538330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2001648247241974, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.010064360592514277, + "grad_norm": 0.08602651953697205, + "kept_token_count": 654.6953125, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9932680428028107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006731951888650656, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3675018310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.15442810058596, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27860, + "valid_token_count": 818.534375 + }, + { + "actual_token_mask_ratio": 0.20013703554868698, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.011555046727880835, + "grad_norm": 0.11132719367742538, + "kept_token_count": 654.4265625, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9933463096618652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066536951810121535, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7598846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.66229553222655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27870, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20016320496797563, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.012200997304171323, + "grad_norm": 0.0630757287144661, + "kept_token_count": 654.4125, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9933740437030792, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006625958625227213, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1235870361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0404968261719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27880, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20037029534578324, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.012744041997939349, + "grad_norm": 0.1267128586769104, + "kept_token_count": 654.003125, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0162, + "mgd_cos_sim": 0.9930271804332733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069728225469589235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.6817687988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3328826904297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27890, + "valid_token_count": 817.88125 + }, + { + "actual_token_mask_ratio": 0.1992047145962715, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.011889574211090803, + "grad_norm": 0.12812477350234985, + "kept_token_count": 655.2984375, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9932823836803436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006717614829540253, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9516082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0058563232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27900, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.199771012365818, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.012953469809144736, + "grad_norm": 0.10098596662282944, + "kept_token_count": 654.60625, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0162, + "mgd_cos_sim": 0.9936038434505463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006396166421473027, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1897247314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.38474426269534, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27910, + "valid_token_count": 818.01875 + }, + { + "actual_token_mask_ratio": 0.20068280398845673, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.013153956318274141, + "grad_norm": 0.10438309609889984, + "kept_token_count": 653.871875, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0165, + "mgd_cos_sim": 0.9933584809303284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006641495227813721, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1515319824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6852966308594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27920, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20040727257728577, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.010402495367452502, + "grad_norm": 0.10986728966236115, + "kept_token_count": 654.2640625, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9931900084018708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006809988431632519, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3920013427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.22838439941404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27930, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.20007405877113343, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.011226813215762377, + "grad_norm": 0.0891089215874672, + "kept_token_count": 654.5671875, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9934691190719604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006530868820846081, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.83623962402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9446594238281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27940, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.20106004774570466, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.010430843289941549, + "grad_norm": 0.13277709484100342, + "kept_token_count": 653.6109375, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9936724662780761, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00632752925157547, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3271453857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6883483886719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27950, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.20014006197452544, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.011307763028889894, + "grad_norm": 0.11168792843818665, + "kept_token_count": 654.3703125, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9935403287410736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006459680665284395, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8677032470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.24090881347655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27960, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.199982750415802, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.010749889258295297, + "grad_norm": 0.10551776736974716, + "kept_token_count": 654.628125, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9927400708198547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007259918376803398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.38869018554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.24903259277346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27970, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.1999024420976639, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.011215856857597828, + "grad_norm": 0.0772724524140358, + "kept_token_count": 654.49375, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9936752259731293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006324772071093321, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.04285888671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1961181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27980, + "valid_token_count": 818.0234375 + }, + { + "actual_token_mask_ratio": 0.19933463186025618, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.012541710073128343, + "grad_norm": 0.08077569305896759, + "kept_token_count": 655.15, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0159, + "mgd_cos_sim": 0.9931879639625549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006812037900090218, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.25579528808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4323974609375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 27990, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.19958680123090744, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.012024103431031108, + "grad_norm": 0.08588679134845734, + "kept_token_count": 655.0234375, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9929893493652344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007010634522885084, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.3544219970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.56719360351565, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28000, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.1999194234609604, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.013181041041389108, + "grad_norm": 0.08046567440032959, + "kept_token_count": 654.6828125, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0164, + "mgd_cos_sim": 0.9935162007808686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006483794003725052, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.97320861816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3140045166016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28010, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.1994974583387375, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.01163447848521173, + "grad_norm": 0.09437394887208939, + "kept_token_count": 655.0109375, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9936588943004608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341098435223102, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.5595947265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0187622070313, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28020, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19939251989126205, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.009495599195361137, + "grad_norm": 0.10274040699005127, + "kept_token_count": 655.1609375, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0128, + "mgd_cos_sim": 0.9934169173240661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006583084166049957, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9302276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.764208984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28030, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.20025536566972732, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.01191931301727891, + "grad_norm": 0.1204385757446289, + "kept_token_count": 654.3625, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9938204228878021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006179577857255936, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.72910461425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1329833984375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28040, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.19950547069311142, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.011912953946739436, + "grad_norm": 0.07255757600069046, + "kept_token_count": 655.0796875, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9934826195240021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006517387926578522, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.222265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.90069274902345, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28050, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.1999841883778572, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.01205153800547123, + "grad_norm": 0.1279294639825821, + "kept_token_count": 654.615625, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9935448467731476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006455150712281465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.27814025878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3983917236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28060, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.199764446914196, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.009544460009783507, + "grad_norm": 0.07944345474243164, + "kept_token_count": 654.9453125, + "learning_rate": 3.385286791820802e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9930931091308594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906904093921184, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8703338623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8219879150391, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28070, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.20017941743135453, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.012284934287890792, + "grad_norm": 0.11271557211875916, + "kept_token_count": 654.275, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9932398974895478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006760091800242662, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9288757324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.4778991699219, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28080, + "valid_token_count": 818.0328125 + }, + { + "actual_token_mask_ratio": 0.20082693696022033, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.01225473741069436, + "grad_norm": 0.1470392346382141, + "kept_token_count": 653.9234375, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9936629474163056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006337049603462219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.641552734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9588592529297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28090, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2006814032793045, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.010606107115745545, + "grad_norm": 0.13097847998142242, + "kept_token_count": 654.0546875, + "learning_rate": 3.281298538383559e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9932213664054871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006778643000870943, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0857238769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0844421386719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28100, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.20041449815034867, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.013812190340831875, + "grad_norm": 0.12072809785604477, + "kept_token_count": 654.2609375, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0174, + "mgd_cos_sim": 0.9929053544998169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007094645313918591, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.10487976074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1701324462891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28110, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.20033399611711503, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.010876605613157153, + "grad_norm": 0.08986987173557281, + "kept_token_count": 654.2640625, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9933875501155853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612429022789001, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.36406860351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5248779296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28120, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.19968319982290267, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.011749215796589852, + "grad_norm": 0.07249567657709122, + "kept_token_count": 655.05625, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9937237560749054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006276250071823597, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.30149536132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83533935546876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28130, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.19994346648454667, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.01202610875479877, + "grad_norm": 0.062069013714790344, + "kept_token_count": 654.7453125, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.993444436788559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006555559951812029, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.80853271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4009582519531, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2010814815759659, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.011604720633476973, + "grad_norm": 0.09213490039110184, + "kept_token_count": 653.7359375, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9930084764957428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006991526391357183, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8363739013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4313507080078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28150, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.19941889345645905, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.011702272016555071, + "grad_norm": 0.0981220230460167, + "kept_token_count": 655.2046875, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9930727660655976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069272351451218125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.22174682617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.13524475097654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28160, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.19997815191745758, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.011788027454167604, + "grad_norm": 0.08127831667661667, + "kept_token_count": 654.5890625, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9930130243301392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006986986007541418, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.89361877441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36647033691406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28170, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.2011873245239258, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.011204988416284322, + "grad_norm": 0.1423691362142563, + "kept_token_count": 653.765625, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9935196995735168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006480299029499292, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3007354736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.92882385253904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28180, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.19948591589927672, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.011812267964705825, + "grad_norm": 0.10136963427066803, + "kept_token_count": 654.803125, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9938644945621491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061354941688477995, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.27251892089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7246124267578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28190, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.19914191961288452, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.00871976655907929, + "grad_norm": 0.08068380504846573, + "kept_token_count": 655.2328125, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9935544490814209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006445574760437012, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4300506591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1509185791016, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28200, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.20025094002485275, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.010101892333477736, + "grad_norm": 0.08234740048646927, + "kept_token_count": 654.471875, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9931248486042022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006875160243362188, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6178466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5075378417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28210, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19976964443922043, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.010104377148672938, + "grad_norm": 0.10257960855960846, + "kept_token_count": 655.2265625, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9935266256332398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064733635634183885, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.70465393066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26401672363284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28220, + "valid_token_count": 818.796875 + }, + { + "actual_token_mask_ratio": 0.19997294545173644, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.008785679191350936, + "grad_norm": 0.08850431442260742, + "kept_token_count": 654.6828125, + "learning_rate": 2.849237816642192e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9935400247573852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064599703066051005, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.1030517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7397125244141, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28230, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.19949295818805696, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.011385230906307697, + "grad_norm": 0.09311093389987946, + "kept_token_count": 655.1328125, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9934244573116302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065755417570471765, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3952117919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.75884399414065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28240, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.1993227109313011, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.011598312202841044, + "grad_norm": 0.09252540022134781, + "kept_token_count": 655.078125, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0149, + "mgd_cos_sim": 0.9933869242668152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613062974065542, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.91709289550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5818786621094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28250, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.19983404725790024, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.012470532674342394, + "grad_norm": 0.08578924089670181, + "kept_token_count": 654.6546875, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9936248660087585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006375144701451063, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.5261535644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5418243408203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28260, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2002256318926811, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.012224251683801413, + "grad_norm": 0.13449294865131378, + "kept_token_count": 654.4921875, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9934320151805878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065679727122187614, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8179077148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48558044433594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28270, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2001280203461647, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.011414172872900964, + "grad_norm": 0.09629388153553009, + "kept_token_count": 654.25625, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.993310523033142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006689488235861063, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.67717895507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3565338134766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28280, + "valid_token_count": 817.953125 + }, + { + "actual_token_mask_ratio": 0.2004646122455597, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.011005591554567217, + "grad_norm": 0.07743000984191895, + "kept_token_count": 654.2359375, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9933097958564758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006690198741853237, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.15216674804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1578369140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28290, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.20038278698921203, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.011100200284272433, + "grad_norm": 0.07519944757223129, + "kept_token_count": 654.221875, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9932312726974487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067687399685382845, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9214111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.67164611816406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28300, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2006770759820938, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.013944727275520563, + "grad_norm": 0.11748415976762772, + "kept_token_count": 654.1921875, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0174, + "mgd_cos_sim": 0.9931747257709503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006825296673923731, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8163818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83746643066405, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28310, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2002516821026802, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.01044563166797161, + "grad_norm": 0.09468840062618256, + "kept_token_count": 654.084375, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9934628367424011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065371435135602955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6618255615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9976440429688, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28320, + "valid_token_count": 817.8625 + }, + { + "actual_token_mask_ratio": 0.20047743618488312, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.010024100914597512, + "grad_norm": 0.12609581649303436, + "kept_token_count": 654.14375, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0134, + "mgd_cos_sim": 0.9932907581329345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006709241773933172, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2620483398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10735473632815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28330, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.20049201399087907, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.011897476436570287, + "grad_norm": 0.1455022692680359, + "kept_token_count": 654.284375, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9929493486881256, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007050638366490602, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.50054321289065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8181640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28340, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.19869657307863237, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.01041852505877614, + "grad_norm": 0.0731833353638649, + "kept_token_count": 655.89375, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9933723926544189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006627627555280924, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.65607299804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4287414550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28350, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.19998950362205506, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.011806119699031115, + "grad_norm": 0.07793363183736801, + "kept_token_count": 654.440625, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9930573284626008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006942679081112147, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5316131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5541534423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28360, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.1992526337504387, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.009997257590293884, + "grad_norm": 0.09901302307844162, + "kept_token_count": 655.0125, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.992928546667099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007071458175778389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.51849975585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7449188232422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28370, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.2000966563820839, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.011420630710199475, + "grad_norm": 0.08020433783531189, + "kept_token_count": 654.325, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9931424021720886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006857607141137123, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.97776794433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.77085266113284, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28380, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.19869506657123565, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.011477895267307758, + "grad_norm": 0.13114920258522034, + "kept_token_count": 655.6796875, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9939625978469848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037419848144055, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7850006103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.67796020507814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28390, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20151740610599517, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.010231614531949162, + "grad_norm": 0.1085694208741188, + "kept_token_count": 653.4921875, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9929236769676208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007076327130198479, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.2463806152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2006469726563, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28400, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.1988767758011818, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.011154073616489768, + "grad_norm": 0.07243887335062027, + "kept_token_count": 655.6203125, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9934439241886139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006556086242198944, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6549774169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.91846923828126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28410, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.1998325541615486, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.009711137507110834, + "grad_norm": 0.07588741183280945, + "kept_token_count": 654.978125, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9935504078865052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449596583843231, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3808166503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4464080810547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28420, + "valid_token_count": 818.55 + }, + { + "actual_token_mask_ratio": 0.2017416775226593, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.010640261741355062, + "grad_norm": 0.08062277734279633, + "kept_token_count": 653.253125, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9935057699680329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064942331053316595, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6450469970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8054931640625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.20006630271673204, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.010098803834989668, + "grad_norm": 0.12235458195209503, + "kept_token_count": 654.70625, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0135, + "mgd_cos_sim": 0.9932560384273529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006743972189724445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1145080566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.12843933105466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28440, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.20113692432641983, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.00998410931788385, + "grad_norm": 0.09856551885604858, + "kept_token_count": 653.8328125, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9934399783611297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006560032349079847, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.4697021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.85010681152346, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.19981170892715455, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.00993293933570385, + "grad_norm": 0.10769595205783844, + "kept_token_count": 654.9796875, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9933518826961517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006648125033825636, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.78028564453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1556854248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28460, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.2006295457482338, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.01161163910292089, + "grad_norm": 0.08793015033006668, + "kept_token_count": 653.9265625, + "learning_rate": 2.131035337397097e-07, + "loss": 0.015, + "mgd_cos_sim": 0.9933039605617523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006696034781634807, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.65517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.21258850097655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28470, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.20010198950767516, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.01190287284553051, + "grad_norm": 0.0949731320142746, + "kept_token_count": 654.5921875, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9933736741542816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006626332830637694, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0485412597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0978302001953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28480, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.19903240948915482, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.00945010078139603, + "grad_norm": 0.09889663010835648, + "kept_token_count": 655.325, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9929389119148254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007061089947819709, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2620056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4693328857422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28490, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19977760314941406, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.01356005584821105, + "grad_norm": 0.1128566563129425, + "kept_token_count": 654.909375, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0172, + "mgd_cos_sim": 0.9927620351314544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007237962633371353, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8517181396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9459594726562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28500, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.20110571384429932, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.010988607024773955, + "grad_norm": 0.07952702790498734, + "kept_token_count": 653.575, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9931852579116821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006814743671566248, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1414855957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.70461120605466, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28510, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.19940511882305145, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.009597620833665133, + "grad_norm": 0.07113835960626602, + "kept_token_count": 655.3046875, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0129, + "mgd_cos_sim": 0.9933499574661255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00665003452450037, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.08131713867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7233917236328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28520, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.20001075267791749, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.009624512307345868, + "grad_norm": 0.07738938927650452, + "kept_token_count": 654.5359375, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9938817143440246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006118280906230211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.08046569824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.097314453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28530, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.20048487931489944, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.010720699187368154, + "grad_norm": 0.09382794052362442, + "kept_token_count": 654.3109375, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9937165021896363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006283500511199236, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.76044311523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.99847717285155, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28540, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2000485271215439, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.01050561056472361, + "grad_norm": 0.07802943140268326, + "kept_token_count": 654.4421875, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.014, + "mgd_cos_sim": 0.9930367529392242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963251251727342, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2146453857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8254821777344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28550, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.20010782033205032, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.009741571079939604, + "grad_norm": 0.1004418432712555, + "kept_token_count": 654.6984375, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0128, + "mgd_cos_sim": 0.9938179612159729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006182038504630327, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.16603088378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.50714111328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28560, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2000685900449753, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.012154371570795774, + "grad_norm": 0.070652537047863, + "kept_token_count": 654.6625, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0155, + "mgd_cos_sim": 0.9933033168315888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006696693692356348, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0963165283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.49391174316406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28570, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.20046123564243318, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.013848096784204244, + "grad_norm": 0.11354222148656845, + "kept_token_count": 654.2109375, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0171, + "mgd_cos_sim": 0.9934642791748047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006535731628537178, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.96904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1623931884766, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28580, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.19896529167890548, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.010506049217656254, + "grad_norm": 0.14325064420700073, + "kept_token_count": 655.4546875, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9934549689292907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00654502036049962, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.4083557128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.24750366210935, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28590, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19975776821374894, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.012247560126706959, + "grad_norm": 0.0852576345205307, + "kept_token_count": 655.0546875, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9931501746177673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00684983292594552, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.38304443359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7499603271484, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28600, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.20052166134119034, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.009878507349640131, + "grad_norm": 0.13276344537734985, + "kept_token_count": 654.3421875, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0131, + "mgd_cos_sim": 0.9934781074523926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006521887052804232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.03004760742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2468627929687, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28610, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.1994113564491272, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.011215620255097748, + "grad_norm": 0.06215490773320198, + "kept_token_count": 655.1453125, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9934216022491456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006578392349183559, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5100830078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.95660095214845, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28620, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.20050276070833206, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.011565653374418616, + "grad_norm": 0.08192113786935806, + "kept_token_count": 654.2859375, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0147, + "mgd_cos_sim": 0.9937528371810913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247162818908691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8913818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7869415283203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28630, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.20000257194042206, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.011728451447561383, + "grad_norm": 0.11937765777111053, + "kept_token_count": 654.6578125, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9930255711078644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006974419951438904, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.70368041992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4734222412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28640, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19990127235651017, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.009697435982525348, + "grad_norm": 0.09352663159370422, + "kept_token_count": 654.53125, + "learning_rate": 1.66027180568904e-07, + "loss": 0.013, + "mgd_cos_sim": 0.9933187246322632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006681269127875566, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1789611816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.45439453125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28650, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.20017847269773484, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.010977614624425768, + "grad_norm": 0.11414044350385666, + "kept_token_count": 654.165625, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.993140584230423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00685941781848669, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5549713134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7180633544922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28660, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.19965024143457413, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.012136335019022226, + "grad_norm": 0.09126141667366028, + "kept_token_count": 654.6796875, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0153, + "mgd_cos_sim": 0.9937332808971405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006266713421791792, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.93861694335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4668701171875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28670, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.19852544665336608, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.008926146058365702, + "grad_norm": 0.09212811291217804, + "kept_token_count": 655.971875, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9935713171958923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006428696122020483, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.2716400146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9878173828125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28680, + "valid_token_count": 818.45625 + }, + { + "actual_token_mask_ratio": 0.20077529549598694, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.011196824023500085, + "grad_norm": 0.09829172492027283, + "kept_token_count": 654.01875, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0145, + "mgd_cos_sim": 0.9933472156524659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006652779597789049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.95830688476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.24982299804685, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28690, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.20021227598190308, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.01096963188610971, + "grad_norm": 0.14907564222812653, + "kept_token_count": 654.596875, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0142, + "mgd_cos_sim": 0.9934545993804932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545415334403515, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.67377014160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3400909423828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28700, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.19859106838703156, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.010432571079581976, + "grad_norm": 0.092811718583107, + "kept_token_count": 655.678125, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.993379408121109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006620581634342671, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.48842163085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.51129150390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28710, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.19989190250635147, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.011988163646310567, + "grad_norm": 0.09313327819108963, + "kept_token_count": 654.7140625, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9930924355983735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006907551363110542, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.62439880371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5173034667969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28720, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.20094502121210098, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.009937439253553748, + "grad_norm": 0.05905555933713913, + "kept_token_count": 654.1734375, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0132, + "mgd_cos_sim": 0.9934665501117707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533460132777691, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8863037109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0928588867188, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28730, + "valid_token_count": 818.6859375 + }, + { + "actual_token_mask_ratio": 0.20107971727848054, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.01063731350004673, + "grad_norm": 0.07664688676595688, + "kept_token_count": 653.7265625, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9936468601226807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006353155430406332, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7001678466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12012939453126, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.20061073154211045, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.012199666164815427, + "grad_norm": 0.0997399315237999, + "kept_token_count": 653.99375, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9935048580169678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00649512903764844, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.12700805664065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3561279296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28750, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.20010405033826828, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.01229500286281109, + "grad_norm": 0.07646137475967407, + "kept_token_count": 654.440625, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0158, + "mgd_cos_sim": 0.9930852353572845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006914754584431648, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4565063476563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0579376220703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28760, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.19960612505674363, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.010483057098463177, + "grad_norm": 0.09775367379188538, + "kept_token_count": 655.1921875, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0139, + "mgd_cos_sim": 0.9932519793510437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006748020928353071, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1667785644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.97240295410154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28770, + "valid_token_count": 818.5828125 + }, + { + "actual_token_mask_ratio": 0.19849249422550203, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.012255471805110574, + "grad_norm": 0.07662436366081238, + "kept_token_count": 655.921875, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9930867493152619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069132460281252864, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.3808135986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5001525878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28780, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.20063704550266265, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.00999901294708252, + "grad_norm": 0.13972754776477814, + "kept_token_count": 654.08125, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0133, + "mgd_cos_sim": 0.9933567106723785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00664328970015049, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.180517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5484222412109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28790, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.19941929280757903, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.01064309454523027, + "grad_norm": 0.10304626077413559, + "kept_token_count": 654.978125, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0141, + "mgd_cos_sim": 0.9930883049964905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006911691185086966, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.82297973632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.86827392578124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28800, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.19965156316757202, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.011938587622717023, + "grad_norm": 0.10100919008255005, + "kept_token_count": 654.8296875, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0152, + "mgd_cos_sim": 0.9935296773910522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006470328941941261, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.33087768554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.27967529296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28810, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.19929785281419754, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.01300691901706159, + "grad_norm": 0.09564843773841858, + "kept_token_count": 655.278125, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0163, + "mgd_cos_sim": 0.9933906376361847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006609369628131389, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1700714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0189910888672, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.19940380305051802, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.012327508442103863, + "grad_norm": 0.10325188934803009, + "kept_token_count": 655.125, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0157, + "mgd_cos_sim": 0.9932758808135986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672411834821105, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.823046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7335968017578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28830, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.19987155348062516, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.011124013736844063, + "grad_norm": 0.11202700436115265, + "kept_token_count": 655.0046875, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9933523714542389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006647621840238571, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.82383117675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8599853515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28840, + "valid_token_count": 818.6203125 + }, + { + "actual_token_mask_ratio": 0.19948411136865615, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.011098065506666898, + "grad_norm": 0.08323315531015396, + "kept_token_count": 654.8859375, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0144, + "mgd_cos_sim": 0.9933367311954499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006663287524133921, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.5737548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9851043701172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28850, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.2004571110010147, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.011577394511550664, + "grad_norm": 0.14195413887500763, + "kept_token_count": 654.275, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0148, + "mgd_cos_sim": 0.9935948133468628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006405195966362953, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9759765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3640594482422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28860, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.19955879151821138, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.014517683628946542, + "grad_norm": 0.13717631995677948, + "kept_token_count": 655.3140625, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.018, + "mgd_cos_sim": 0.9929844737052917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007015522662550211, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2558319091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.52581481933595, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28870, + "valid_token_count": 818.6953125 + }, + { + "actual_token_mask_ratio": 0.19978393018245696, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.011722622672095895, + "grad_norm": 0.11667081713676453, + "kept_token_count": 655.003125, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9931736409664154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068263615481555465, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5211944580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.855078125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28880, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.19966494292020798, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.010250378819182514, + "grad_norm": 0.100735142827034, + "kept_token_count": 654.775, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0136, + "mgd_cos_sim": 0.9933829247951508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617080327123404, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8596221923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.32275390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28890, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.2005078062415123, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.01210107672959566, + "grad_norm": 0.09133922308683395, + "kept_token_count": 654.1375, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0154, + "mgd_cos_sim": 0.9933737695217133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006626241281628609, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2842987060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.756787109375, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28900, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.20068217515945436, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.012938656844198704, + "grad_norm": 0.10677706450223923, + "kept_token_count": 654.1609375, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0163, + "mgd_cos_sim": 0.9933605670928956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639434397220612, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.27960205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8168212890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28910, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.19908436387777328, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.010570075735449791, + "grad_norm": 0.10270417481660843, + "kept_token_count": 655.415625, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0137, + "mgd_cos_sim": 0.9937286674976349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006271337438374758, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.2549774169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2093902587891, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28920, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.19916384369134904, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.011778693180531263, + "grad_norm": 0.17771963775157928, + "kept_token_count": 655.2515625, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0151, + "mgd_cos_sim": 0.9932745933532715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725406274199486, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1966491699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.46551818847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28930, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20115203112363816, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.010519181843847036, + "grad_norm": 0.08666183054447174, + "kept_token_count": 653.7015625, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0138, + "mgd_cos_sim": 0.9934589505195618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006541057303547859, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.27112731933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2656280517578, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28940, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2003423124551773, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.013954282412305475, + "grad_norm": 0.059694673866033554, + "kept_token_count": 654.3609375, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0173, + "mgd_cos_sim": 0.9932304620742798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006769532151520252, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8894958496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.56541442871094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28950, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.19941062480211258, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.01252890434116125, + "grad_norm": 0.08147720247507095, + "kept_token_count": 654.859375, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0157, + "mgd_cos_sim": 0.9935594499111176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006440552975982427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8952972412109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9890869140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28960, + "valid_token_count": 817.971875 + }, + { + "actual_token_mask_ratio": 0.19997092485427856, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.010765486257150769, + "grad_norm": 0.06445399671792984, + "kept_token_count": 654.7078125, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9932235777378082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006776437908411026, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9146209716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0371032714844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28970, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.19911144226789473, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.011632416397333145, + "grad_norm": 0.11824371665716171, + "kept_token_count": 655.1734375, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0149, + "mgd_cos_sim": 0.9933760821819305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006623921915888786, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6148284912109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28980, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.20070813447237015, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.010197939444333315, + "grad_norm": 0.08018897473812103, + "kept_token_count": 653.740625, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0138, + "mgd_cos_sim": 0.9927719473838806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007228038087487221, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7269012451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3004516601562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 28990, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.1995893746614456, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.010227428376674652, + "grad_norm": 0.06536674499511719, + "kept_token_count": 654.7375, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0133, + "mgd_cos_sim": 0.9937634587287902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062365435063838955, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.73756713867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0858123779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29000, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.19928018301725386, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.011276051541790367, + "grad_norm": 0.09334469586610794, + "kept_token_count": 654.9625, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.993130910396576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006869099009782076, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9788299560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1436828613281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29010, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.19985318481922149, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.012260094331577421, + "grad_norm": 0.1246677115559578, + "kept_token_count": 654.9578125, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0157, + "mgd_cos_sim": 0.9931359171867371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006864080019295215, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.0946929931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3585723876953, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29020, + "valid_token_count": 818.5453125 + }, + { + "actual_token_mask_ratio": 0.20097400099039078, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.011026176018640399, + "grad_norm": 0.09686118364334106, + "kept_token_count": 653.434375, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9930725157260895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006927493866533041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8600616455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8269378662109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29030, + "valid_token_count": 817.79375 + }, + { + "actual_token_mask_ratio": 0.20026398301124573, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.011273604817688465, + "grad_norm": 0.1189875602722168, + "kept_token_count": 654.234375, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9935369610786438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064630330540239814, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5492401123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3761444091797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29040, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.19944292604923247, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.012479585502296687, + "grad_norm": 0.06520625203847885, + "kept_token_count": 655.275, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0159, + "mgd_cos_sim": 0.9931553781032563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006844619382172823, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.98492126464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.45130615234376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29050, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.20015751123428344, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.01221980769187212, + "grad_norm": 0.08723606169223785, + "kept_token_count": 654.246875, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.993315064907074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006684930622577667, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.30950622558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.662744140625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29060, + "valid_token_count": 817.9671875 + }, + { + "actual_token_mask_ratio": 0.2003561496734619, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.01201891927048564, + "grad_norm": 0.12273538112640381, + "kept_token_count": 654.265625, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9930840790271759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006915933173149824, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.47364196777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.59288635253904, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29070, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.19940511733293534, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.010773473232984544, + "grad_norm": 0.09717315435409546, + "kept_token_count": 655.2984375, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9928679883480072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007132012955844402, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3366668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0804138183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29080, + "valid_token_count": 818.51875 + }, + { + "actual_token_mask_ratio": 0.19995491206645966, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.009852809365838766, + "grad_norm": 0.08603012561798096, + "kept_token_count": 654.590625, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0131, + "mgd_cos_sim": 0.9935534417629241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006446561310440302, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.32062377929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3859008789062, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29090, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.20092494189739227, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.011164459446445108, + "grad_norm": 0.06891617923974991, + "kept_token_count": 653.59375, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9934693038463592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006530680973082781, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8575927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65248413085936, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29100, + "valid_token_count": 817.9328125 + }, + { + "actual_token_mask_ratio": 0.2001967579126358, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.011725705675780773, + "grad_norm": 0.11459250003099442, + "kept_token_count": 654.6984375, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0151, + "mgd_cos_sim": 0.993224424123764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006775575969368219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0452606201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8973693847656, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29110, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.1994485542178154, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.014215216645970941, + "grad_norm": 0.1319877803325653, + "kept_token_count": 654.978125, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0177, + "mgd_cos_sim": 0.992975652217865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007024356815963984, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1685791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4034027099609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29120, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.19984905868768693, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.012234929576516151, + "grad_norm": 0.08642677962779999, + "kept_token_count": 654.7109375, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0157, + "mgd_cos_sim": 0.9929933488368988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070066522806882855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.81101379394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.39299926757815, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29130, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.20059711933135987, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.012507663387805224, + "grad_norm": 0.07480547577142715, + "kept_token_count": 654.1203125, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0158, + "mgd_cos_sim": 0.993398004770279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00660199411213398, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4494934082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10204162597654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29140, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.19985058009624482, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.01163631328381598, + "grad_norm": 0.07787119597196579, + "kept_token_count": 654.8046875, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0152, + "mgd_cos_sim": 0.9928769171237946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007123092003166675, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.10924072265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7314025878906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29150, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.1996169075369835, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.011200433829799294, + "grad_norm": 0.09272102266550064, + "kept_token_count": 654.75, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9937121152877808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006287885922938585, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8468719482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5682739257812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29160, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.19959538727998732, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.011165974335744977, + "grad_norm": 0.07501880824565887, + "kept_token_count": 654.9078125, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9934260368347168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006573967076838017, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.00453186035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2560607910156, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29170, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.20031939297914506, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.01277728397399187, + "grad_norm": 0.10486206412315369, + "kept_token_count": 654.3921875, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0161, + "mgd_cos_sim": 0.993309885263443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066901084035635, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.70924682617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.11524963378906, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29180, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.19939387887716292, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.010657545970752835, + "grad_norm": 0.09543885290622711, + "kept_token_count": 655.2125, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9930742025375366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006925792340189219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.20786743164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6071044921875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29190, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.200377456843853, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.009291645884513856, + "grad_norm": 0.08534328639507294, + "kept_token_count": 654.1984375, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0125, + "mgd_cos_sim": 0.9936069965362548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006393005233258009, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.084423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6529998779297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29200, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.20044352263212203, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.011079497495666146, + "grad_norm": 0.14473716914653778, + "kept_token_count": 654.4046875, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9937636613845825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006236327346414328, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0017944335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7895782470703, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29210, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.1994316339492798, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.011634089797735215, + "grad_norm": 0.09057946503162384, + "kept_token_count": 655.096875, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0148, + "mgd_cos_sim": 0.9936189532279969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00638106195256114, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6558837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2904296875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29220, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.2019846633076668, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.011098860809579492, + "grad_norm": 0.09703832864761353, + "kept_token_count": 652.94375, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9934816122055053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006518392264842987, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4165802001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8365173339844, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20009153932332993, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.011359693249687552, + "grad_norm": 0.06899625808000565, + "kept_token_count": 655.0609375, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9932409346103668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006759059894829988, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.07178955078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7926330566406, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29240, + "valid_token_count": 818.9203125 + }, + { + "actual_token_mask_ratio": 0.20045326948165892, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.010771260596811771, + "grad_norm": 0.08716921508312225, + "kept_token_count": 654.2453125, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.9933565139770508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00664348965510726, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.11191711425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9305389404297, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29250, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20000623166561127, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.01208994328044355, + "grad_norm": 0.09449019283056259, + "kept_token_count": 654.65, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0153, + "mgd_cos_sim": 0.9935186505317688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006481353845447302, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.93638000488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.06162719726564, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29260, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.19953180104494095, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.0120984326582402, + "grad_norm": 0.10907699912786484, + "kept_token_count": 654.9515625, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0154, + "mgd_cos_sim": 0.9934892356395721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006510756630450487, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.3723999023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.631103515625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29270, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.20029299557209015, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.01169869052246213, + "grad_norm": 0.12945249676704407, + "kept_token_count": 654.4375, + "learning_rate": 4.734926068397038e-08, + "loss": 0.015, + "mgd_cos_sim": 0.99333336353302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066666420549154285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.02480163574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7919555664063, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29280, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.19937173575162886, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.012011507665738464, + "grad_norm": 0.11310695856809616, + "kept_token_count": 654.8640625, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0154, + "mgd_cos_sim": 0.9931268811225891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006873132195323706, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.94081115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7867156982422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29290, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.20000457912683486, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.010579579090699554, + "grad_norm": 0.12862205505371094, + "kept_token_count": 654.4890625, + "learning_rate": 4.476011737867658e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9931110322475434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006888971664011478, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.01776123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.92775268554686, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29300, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.1992769256234169, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.010858258651569485, + "grad_norm": 0.08825144916772842, + "kept_token_count": 655.3390625, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.9928617954254151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007138199079781771, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5312744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9934417724609, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29310, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.20036637634038926, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.013168798759579659, + "grad_norm": 0.11867590248584747, + "kept_token_count": 654.3015625, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0164, + "mgd_cos_sim": 0.9934669494628906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533037964254617, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.317041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3589691162109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29320, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2004009246826172, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.010044579626992345, + "grad_norm": 0.08844231814146042, + "kept_token_count": 654.2625, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0134, + "mgd_cos_sim": 0.9932958245277405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067041777074337006, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.68024291992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.50030517578125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29330, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2012641727924347, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.01224463335238397, + "grad_norm": 0.09720256924629211, + "kept_token_count": 653.76875, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0155, + "mgd_cos_sim": 0.9934413135051727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006558691523969174, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1286224365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4717987060547, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29340, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.19967800974845887, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.012673525884747506, + "grad_norm": 0.0807732418179512, + "kept_token_count": 654.9234375, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0159, + "mgd_cos_sim": 0.9935818135738372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418195087462664, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.79493103027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.87343444824216, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29350, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.19974273592233657, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.011297867400571704, + "grad_norm": 0.08613205701112747, + "kept_token_count": 654.7375, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0147, + "mgd_cos_sim": 0.9932013988494873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067986039444804195, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.50397338867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2053741455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29360, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.1992662400007248, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.011169632337987423, + "grad_norm": 0.11295747011899948, + "kept_token_count": 655.5171875, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9931270837783813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006872915010899305, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.393408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9381164550781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29370, + "valid_token_count": 818.6421875 + }, + { + "actual_token_mask_ratio": 0.19983756840229033, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.010501730162650347, + "grad_norm": 0.10010130703449249, + "kept_token_count": 654.878125, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.9932837128639221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006716285366564989, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.89881896972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0654724121094, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29380, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.20033114403486252, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.010993169108405709, + "grad_norm": 0.09819605946540833, + "kept_token_count": 654.19375, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0144, + "mgd_cos_sim": 0.993181586265564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006818411871790886, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.65021362304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.98016967773435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29390, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.19968907684087753, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.010999905783683062, + "grad_norm": 0.10909450799226761, + "kept_token_count": 654.81875, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9935015738010406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006498414743691683, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0001647949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3833038330078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29400, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.20031443238258362, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.010795491887256503, + "grad_norm": 0.06872088462114334, + "kept_token_count": 654.3984375, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0139, + "mgd_cos_sim": 0.9937804222106934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006219578813761473, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.45982666015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2741241455078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29410, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.20069077610969543, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.011268771439790725, + "grad_norm": 0.145293191075325, + "kept_token_count": 653.9296875, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9933190107345581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006680990569293499, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4260986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5104461669922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29420, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.20076781064271926, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.014548024442046881, + "grad_norm": 0.13021038472652435, + "kept_token_count": 654.2765625, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.018, + "mgd_cos_sim": 0.9930904924869537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006909510400146246, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2106872558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7989898681641, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29430, + "valid_token_count": 818.6328125 + }, + { + "actual_token_mask_ratio": 0.19945310950279235, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.008189154928550125, + "grad_norm": 0.06097909063100815, + "kept_token_count": 654.8796875, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0115, + "mgd_cos_sim": 0.9933087944984436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006691213417798281, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.51806640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6264587402344, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29440, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.20009507536888121, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.01017835340462625, + "grad_norm": 0.08834347128868103, + "kept_token_count": 654.55, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0134, + "mgd_cos_sim": 0.9935814678668976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418532785028219, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0514770507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3121765136719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29450, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.19977626502513884, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.013761642575263976, + "grad_norm": 0.16852176189422607, + "kept_token_count": 654.8828125, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0172, + "mgd_cos_sim": 0.9930394291877747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069605506025254725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1854675292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9332763671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29460, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2002209410071373, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.010881718667224049, + "grad_norm": 0.07865441590547562, + "kept_token_count": 654.4671875, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0143, + "mgd_cos_sim": 0.9932577848434448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006742212269455194, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.97496643066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.6716705322266, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29470, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.1996512681245804, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.01112142689526081, + "grad_norm": 0.10256456583738327, + "kept_token_count": 654.803125, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0146, + "mgd_cos_sim": 0.9929839372634888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007016064506024123, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0170379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.726513671875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29480, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19994774460792542, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.012344747129827739, + "grad_norm": 0.08979484438896179, + "kept_token_count": 654.703125, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0157, + "mgd_cos_sim": 0.9932167053222656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006783291138708592, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.22528381347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7684692382812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29490, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.20008465200662612, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.0119274556171149, + "grad_norm": 0.09958748519420624, + "kept_token_count": 654.384375, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.9926072955131531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007392708025872707, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.24879150390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.59703674316404, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29500, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.19968734979629515, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.011051636841148137, + "grad_norm": 0.10763320326805115, + "kept_token_count": 655.0515625, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0142, + "mgd_cos_sim": 0.9937394022941589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260601617395878, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.23459777832034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.81317443847655, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29510, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.20029281526803971, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.011519751604646444, + "grad_norm": 0.08910799771547318, + "kept_token_count": 654.0765625, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.015, + "mgd_cos_sim": 0.9930345296859742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069654684513807295, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3716125488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0301574707031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29520, + "valid_token_count": 817.89375 + }, + { + "actual_token_mask_ratio": 0.19938593208789826, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.010549570014700294, + "grad_norm": 0.052101291716098785, + "kept_token_count": 654.978125, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0138, + "mgd_cos_sim": 0.9935109496116639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064890551380813125, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.00635986328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8697113037109, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29530, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.20003409832715988, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.01167124449275434, + "grad_norm": 0.08095454424619675, + "kept_token_count": 654.521875, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.015, + "mgd_cos_sim": 0.9932857692241669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006714219227433205, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5611175537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9950317382812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29540, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.1992480531334877, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.01014138339087367, + "grad_norm": 0.10327477008104324, + "kept_token_count": 655.3234375, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0137, + "mgd_cos_sim": 0.992877197265625, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007122785318642855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1205322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.31122436523435, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29550, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.1993898406624794, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.011535999923944473, + "grad_norm": 0.09506121277809143, + "kept_token_count": 655.1515625, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0149, + "mgd_cos_sim": 0.9931810200214386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006818986032158136, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2820526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.29210205078124, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29560, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.19984933882951736, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.010177197586745023, + "grad_norm": 0.08721014112234116, + "kept_token_count": 654.6859375, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0135, + "mgd_cos_sim": 0.9934109449386597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065890542231500145, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.7746337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8541687011719, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29570, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2005848467350006, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.013681408390402794, + "grad_norm": 0.10508638620376587, + "kept_token_count": 654.2484375, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0169, + "mgd_cos_sim": 0.9935586988925934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006441297195851803, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0213928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7945922851562, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29580, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.19958065301179886, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.010683267284184695, + "grad_norm": 0.13135240972042084, + "kept_token_count": 655.0375, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9933171272277832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006682868674397468, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.38987426757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5447601318359, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29590, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.19964759349822997, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.011114716995507479, + "grad_norm": 0.09662230312824249, + "kept_token_count": 654.75625, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0145, + "mgd_cos_sim": 0.9931540012359619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006845999602228403, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.37681884765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5809600830078, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29600, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.20068635642528534, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.010512861562892795, + "grad_norm": 0.12272530049085617, + "kept_token_count": 654.15625, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.014, + "mgd_cos_sim": 0.9930873394012452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006912640482187271, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8113189697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.7463104248047, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29610, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2015342652797699, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.01065107686445117, + "grad_norm": 0.07819013297557831, + "kept_token_count": 653.3640625, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0141, + "mgd_cos_sim": 0.993021547794342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00697844959795475, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.19720153808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7948059082031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29620, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.19985361099243165, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.012298841867595911, + "grad_norm": 0.09481173008680344, + "kept_token_count": 654.68125, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.9934368133544922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065631774254143235, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.90135803222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8136962890625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29630, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.1999606817960739, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.012303864443674683, + "grad_norm": 0.10692746937274933, + "kept_token_count": 654.546875, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0156, + "mgd_cos_sim": 0.9934912085533142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065087912604212764, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.61163940429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.05638732910154, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29640, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2007488116621971, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.011462382925674319, + "grad_norm": 0.10024628788232803, + "kept_token_count": 654.1265625, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0148, + "mgd_cos_sim": 0.9932964682579041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006703542266041041, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.55931396484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2572845458984, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29650, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.2001424714922905, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.010584220662713051, + "grad_norm": 0.0955168604850769, + "kept_token_count": 654.621875, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0137, + "mgd_cos_sim": 0.9938401222229004, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159865949302912, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.65203857421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.08408203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29660, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.20004759579896927, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.009236358059570193, + "grad_norm": 0.09312025457620621, + "kept_token_count": 654.678125, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0125, + "mgd_cos_sim": 0.9935170352458954, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006482955720275641, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.71454162597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58518676757814, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29670, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.20036476403474807, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.01132427784614265, + "grad_norm": 0.1000494658946991, + "kept_token_count": 654.328125, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0147, + "mgd_cos_sim": 0.9932944357395173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006705552712082863, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1531494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0109100341797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29680, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2006229728460312, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.012257223250344396, + "grad_norm": 0.0833323523402214, + "kept_token_count": 654.0953125, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0156, + "mgd_cos_sim": 0.9934135913848877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006586389522999525, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1761505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4549102783203, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29690, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.20003726929426194, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.01250264928676188, + "grad_norm": 0.1365537941455841, + "kept_token_count": 654.6796875, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0159, + "mgd_cos_sim": 0.9931766808032989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006823324225842952, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2383605957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1794128417969, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29700, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.20060452371835708, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.01090114088729024, + "grad_norm": 0.09140040725469589, + "kept_token_count": 654.0859375, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0143, + "mgd_cos_sim": 0.993143516778946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068564900197088715, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.51263732910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4891723632812, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29710, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.19975774586200715, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.010290684876963496, + "grad_norm": 0.07590296864509583, + "kept_token_count": 654.834375, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0134, + "mgd_cos_sim": 0.9937200307846069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062799714505672455, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1036010742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0941925048828, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29720, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.199393630027771, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.01088323425501585, + "grad_norm": 0.11350550502538681, + "kept_token_count": 655.1375, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0141, + "mgd_cos_sim": 0.9935815691947937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418436206877232, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.19585876464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.06109008789065, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.20012032836675644, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.009582894202321769, + "grad_norm": 0.1019514948129654, + "kept_token_count": 654.734375, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0128, + "mgd_cos_sim": 0.9936350286006927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006364964414387941, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1607330322266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9710266113281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29740, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.19973817169666291, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.012022798974066972, + "grad_norm": 0.11662832647562027, + "kept_token_count": 655.0390625, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0154, + "mgd_cos_sim": 0.9931778490543366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006822140607982874, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2333740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.84657592773436, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29750, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.19973749816417694, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.011825432023033499, + "grad_norm": 0.0882391482591629, + "kept_token_count": 654.91875, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0151, + "mgd_cos_sim": 0.9933513164520263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006648678705096245, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.54883728027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5958709716797, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29760, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.20034096986055375, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.01178110633045435, + "grad_norm": 0.10054934769868851, + "kept_token_count": 654.3140625, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0152, + "mgd_cos_sim": 0.993201631307602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067983769811689855, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8266998291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.47829284667966, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29770, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.19952317029237748, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.01264860057272017, + "grad_norm": 0.0989808738231659, + "kept_token_count": 655.028125, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0161, + "mgd_cos_sim": 0.993111914396286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068880804814398285, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1112915039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0073760986328, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29780, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.19909430146217347, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.010897500347346068, + "grad_norm": 0.0806790217757225, + "kept_token_count": 655.275, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0142, + "mgd_cos_sim": 0.9934352219104767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006564770266413689, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.91307373046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1926025390625, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29790, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.20066787898540497, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.011072435230016709, + "grad_norm": 0.09219235926866531, + "kept_token_count": 653.8359375, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0144, + "mgd_cos_sim": 0.993441390991211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006558605656027794, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7421630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2764465332031, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29800, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.2005166843533516, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.011915838159620763, + "grad_norm": 0.09760432690382004, + "kept_token_count": 654.175, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0153, + "mgd_cos_sim": 0.9932762205600738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672377496957779, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.00169982910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.14422607421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29810, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.1995244190096855, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.011135239899158478, + "grad_norm": 0.10446811467409134, + "kept_token_count": 654.85625, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0145, + "mgd_cos_sim": 0.9932996153831481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067003917880356315, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6704559326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.002783203125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29820, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.19973715096712114, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.011779073718935251, + "grad_norm": 0.10439015179872513, + "kept_token_count": 654.9046875, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0149, + "mgd_cos_sim": 0.9938246309757233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006175378896296024, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1922546386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9994262695312, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29830, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.1997616782784462, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.010745199071243406, + "grad_norm": 0.07206352800130844, + "kept_token_count": 654.734375, + "learning_rate": 2.362172677446428e-09, + "loss": 0.014, + "mgd_cos_sim": 0.9934560537338257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065439285710453985, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2819885253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9987274169922, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29840, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.19844749867916106, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.012167470902204514, + "grad_norm": 0.08978695422410965, + "kept_token_count": 655.7875, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0156, + "mgd_cos_sim": 0.9931945264339447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006805469561368227, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1101440429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.9745361328125, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29850, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.19994972348213197, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.012573026679456234, + "grad_norm": 0.07006334513425827, + "kept_token_count": 654.8671875, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0156, + "mgd_cos_sim": 0.9938996732234955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006100316345691681, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.14482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6402648925781, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29860, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.19987280666828156, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.01087777391076088, + "grad_norm": 0.09436020255088806, + "kept_token_count": 654.59375, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0141, + "mgd_cos_sim": 0.993467491865158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006532516516745091, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5652618408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2303131103516, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29870, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2007504239678383, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.011475749174132943, + "grad_norm": 0.10052110254764557, + "kept_token_count": 653.909375, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0148, + "mgd_cos_sim": 0.993445235490799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006554761528968811, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.7613952636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.40562133789064, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29880, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.1999206930398941, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.012636417895555497, + "grad_norm": 0.09781546145677567, + "kept_token_count": 654.625, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0159, + "mgd_cos_sim": 0.9935668170452118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433171220123768, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1672821044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.62522888183594, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29890, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.19952546209096908, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.011299238447099924, + "grad_norm": 0.07036139070987701, + "kept_token_count": 655.1875, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0147, + "mgd_cos_sim": 0.9932128071784974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006787196360528469, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3519561767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.15331115722654, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29900, + "valid_token_count": 818.4984375 + }, + { + "actual_token_mask_ratio": 0.20037119090557098, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.010453490540385246, + "grad_norm": 0.07258306443691254, + "kept_token_count": 654.4859375, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0137, + "mgd_cos_sim": 0.9934792339801788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520775519311428, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.33582763671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.40003662109376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29910, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.20046280771493913, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.011429035803303123, + "grad_norm": 0.07961851358413696, + "kept_token_count": 654.125, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0147, + "mgd_cos_sim": 0.993434488773346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006565517652779818, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6098602294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8394500732422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29920, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.19974290877580642, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.014355114242061973, + "grad_norm": 0.18825535476207733, + "kept_token_count": 654.7296875, + "learning_rate": 4.593944767161684e-10, + "loss": 0.018, + "mgd_cos_sim": 0.9927213549613952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072786473669111725, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1000030517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9268859863281, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29930, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.19998033493757247, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.010995063418522476, + "grad_norm": 0.08295963704586029, + "kept_token_count": 654.6453125, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0144, + "mgd_cos_sim": 0.9932600438594819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067399512976408, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2495147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.66761474609376, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29940, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2009039044380188, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.0121329833753407, + "grad_norm": 0.1213517040014267, + "kept_token_count": 653.8765625, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0155, + "mgd_cos_sim": 0.9932486951351166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006751294154673815, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0391418457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.74451904296876, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29950, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.20019121170043946, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.010821570735424757, + "grad_norm": 0.0968388020992279, + "kept_token_count": 654.4375, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0141, + "mgd_cos_sim": 0.993379431962967, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006620579399168491, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5077728271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4384735107422, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29960, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.19987283051013946, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.01095895110629499, + "grad_norm": 0.10085541754961014, + "kept_token_count": 654.5453125, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0142, + "mgd_cos_sim": 0.9935525178909301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064475004561245445, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.390576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6190887451172, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29970, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.19983357340097427, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.010081629222258925, + "grad_norm": 0.12000854313373566, + "kept_token_count": 654.534375, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0134, + "mgd_cos_sim": 0.9934055089950562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065944895148277284, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7410034179687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.34107666015626, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29980, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.20046470165252686, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.009969734121114016, + "grad_norm": 0.09277121722698212, + "kept_token_count": 654.375, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.013, + "mgd_cos_sim": 0.9938431739807129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156817357987165, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1240478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9801544189453, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 29990, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.1998548239469528, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.00966070182621479, + "grad_norm": 0.08414877206087112, + "kept_token_count": 654.971875, + "learning_rate": 9.113208188082922e-14, + "loss": 0.013, + "mgd_cos_sim": 0.993403434753418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00659656822681427, + "mgd_loss_weight": 0.5, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8854187011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.791357421875, + "mgd_token_mask_ratio_cfg": 0.20000000298023224, + "step": 30000, + "valid_token_count": 818.5625 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.018889050210515658, + "train_runtime": 57157.188, + "train_samples_per_second": 67.183, + "train_steps_per_second": 0.525 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/training_args.bin b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..bfc7acee4053869ff080e880d72d67fe3f773f4d --- /dev/null +++ b/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_ablation/mgd_token_mask_ratio_0.2/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d883c3803a3d226570b26fe5fa0f00c7b040bc7c0e9b5e1a6955f80ad5424ab +size 5969